Rolling ability scores is a time-honored tradition across many editions of D&D. However, it can sometimes cause problems for players and/or the DM. For example, one player character may end up much weaker or much stronger than the rest of the party, which can result in a poor experience for some of the players. In other cases, a player may have their characters repeatedly commit suicide-by-monster so they can try to reroll for higher stats, which can be quite frustrating for the GM and other players.

A lot of character power is dependent on ability scores. How much time a player gets in the spotlight can be heavily impacted by how powerful their character is. Having one character be much more/less powerful than the rest of the party can result in imbalanced time for the player of that character in the spotlight, which results in players having less fun. Giving all PCs fairly equivalent ability scores can help avoid that problem.

What approaches are available to mitigate these problems?

Note: Answers should ideally be able to prevent both the "Joe rolled all 7s, and his character is useless" problem and the "Karen rolled all 18s and her character makes everyone else's character useless" problem. That is, an answer that only avoids very low average/total scores is not as good as one that avoids both very low and very high average/total scores.

I intended this to be a canonical question for all editions. I don't mind edition-specific answers as long as they're not also character-specific.

  • 1
    \$\begingroup\$ You mention what problems rolling has for you, several answers came up about non-random stat generation but you accepted none of them, so we're missing part of your issue here. Why exactly do you like to roll? What is it that rolling has and other methods are missing that you like? \$\endgroup\$
    – Rad80
    Commented Feb 8, 2023 at 8:51
  • 1
    \$\begingroup\$ @Rad80 I personally do not like to roll. But lots of other people like to roll, or had no idea that there were options other than rolling, because D&D rulebooks have presented rolling ability scores as the default, or only, option in almost every edition. D&D has taught a lot of people that rolling ability scores is the "correct" way of generating them. \$\endgroup\$
    – Oblivious Sage
    Commented Feb 8, 2023 at 13:46

25 Answers 25


Option One: Non-Random Ability Score Generation

Rather than giving up control of how powerful player characters are to the whims of fate, you can instead use systems that attempt to consistently produce ability scores at an established level of quality. The two most popular schemes for doing so are standard arrays and point-buy systems.

Standard arrays present players with between one and three pre-generated sets of ability scores. Players select a set and assign the scores within to their abilities however they want. For example, players may be given two arrays: 16 14 14 13 10 8 and 16 14 13 12 11 10. Carol wants to make a fighter, and chooses the second array. She then picks the 16 for her strength, the 14 for her constitution, the 13 for her dexterity, the 12 for her wisdom, the 11 for her intelligence, and the 10 for her charisma. Bob wants to make a wizard, so he chooses the first array, and assigns the 16 to his Int, the 14s to his Dex and Con, the 13 to his Cha, the 10 to his Wis, and the 8 to his Str.

Point-buy systems instead give the player some sort of base ability scores and then a pool of points they can spend to increase ability scores. In some systems each point you spend increases an ability score by 1, while in others getting a single high score costs more than getting two medium scores. The Pathfinder point-buy scheme is an example of the latter. Point-buy systems give the players more control over their characters' ability scores, which generally makes them happier, but it can also increase the potential/temptation of heavily min-maxing characters (e.g. a fighter with 18s in Str/Dex/Con and 6s in Int/Wis/Cha), which can still result in some frustration for the DM.

Option Two: Collective Ability Score Generation

Most problems that arise from rolling ability scores ultimately center around large differences between player characters. We can resolve this issue, and still roll ability scores, by stealing the idea of having a set of arrays from the first option. Instead of each player rolling an array of ability scores for their own character, each player rolls an array of ability scores for the party and then chooses any of the rolled arrays to use for their character. Once generated, these arrays should be saved and used again for any characters created later, rather than rolling an additional array.

For example, Alex, Betty, and Charles are making character's for Dana's new campaign. Alex rolls 6 11 8 13 9 10, Betty rolls 18 7 12 11 15 10, and Charles rolls 16 12 14 13 15 9. Alex and Charles decide to use the array that Betty rolled, so they can put the 18 in their main ability score, while Betty opts to use the array Charles rolled so she can make a more MAD (Multiple Attribute Dependent) character. When Alex's character dies a few levels in, he doesn't roll a new set of ability scores for his new character. Instead, he goes back to the three arrays generated when the campaign started and chooses one of them. If Eric joins them a few levels after that, he also would use one of the same arrays everyone else picked from when they created a character.

This approach means that if a single player rolls poorly, they're not stuck with a weaker character. If a single player rolls well, everyone can make characters who are just as strong. While this approach tends to slightly increase average party strength, having all the characters on an even playing field makes it easier for the DM to adjust encounters appropriately.

  • 12
    \$\begingroup\$ I wish this also included some discussion of in-play ways to make "uneven ability scores" =/= "a problem for the table." It seems to assume that assuring parity is the way to avoid problems, but in my experience that's not the only way. \$\endgroup\$
    – nitsua60
    Commented Oct 10, 2018 at 19:49
  • 5
    \$\begingroup\$ @nitsua60 Assuring parity is a broadly applicable way of avoiding the problem. There are certainly other ways to deal with the resulting problems, but there's a big difference between, "We're playing 4e and Joe's barbarian's highest stat is a 13, what can we do," and, "We're playing 3.5e and Kate's druid's lowest stat is a 16, what can we do," when you're discussing those alternate approaches. Assuring parity in rolls ahead of time is a solution that can work for almost everyone. \$\endgroup\$
    – Oblivious Sage
    Commented Oct 10, 2018 at 20:04
  • 1
    \$\begingroup\$ Can you please explain the meaning of the MAD acronym? I don't recall encountering it before. \$\endgroup\$
    – Beofett
    Commented Oct 12, 2018 at 16:32
  • 11
    \$\begingroup\$ @Beofett MAD = Multiple Attribute Dependent. For example, a 3.5 monk who needs Str for attack & damage bonuses, Dex for AC, Con for hit points, and Wis for their abilities. \$\endgroup\$
    – Oblivious Sage
    Commented Oct 12, 2018 at 17:29

Draft Ability Scores

One method I've used in the past is to create a pool of ability scores from all the players' rolls. Then each player (order decided by dice roll) chooses an Ability Score from the pool. This means that each player will get some high numbers and some low numbers.

The draft order is important as the first roller will get higher on average with just a rotation so I have the last player choose their first and second scores consecutively and then each even numbered score has an inverted pick order.


Bob rolls 8, 12, 12, 13, 14, 14
Mark rolls 12, 14, 15, 16, 16, 17
Anne rolls 9, 11, 12, 15, 15, 16

These all go in a pool and the first pick order is Mark, Anne, Bob. As stated, first players choose in order, then in reverse order, then in forward order again and so on.

  • 1st pick: Mark 17, Anne 16, Bob 16
  • 2nd pick: Bob 16, Anne 15, Mark 15
  • 3rd pick: Mark 15, Anne 14, Bob 14
  • 4th pick: Bob 14, Anne 13, Mark 12
  • 5th pick: Mark 12, Anne 12, Bob 12
  • 6th pick: Bob: 11, Anne 9, Mark 8

The end result is:

Mark: 17, 15, 15, 12, 12, 8
Anne: 16, 15, 14, 13, 12, 9
Bob: 16, 16, 14, 14, 12, 11

The last pick ends up with the most middling result but they are all fairly balanced to each other.

  • 1
    \$\begingroup\$ I can attest to this being very effective. Came up with a system just like this over the summer and it's only drawback is that if someone dies (or something) you have to figure that out in some other fashion. \$\endgroup\$
    – blurry
    Commented Oct 11, 2018 at 14:08
  • 3
    \$\begingroup\$ This goes some way to solving OP's problems, but introduces a new problem that the first person to pick has an advantage. Related question: rpg.stackexchange.com/q/120334/36002 \$\endgroup\$
    – Richard
    Commented Oct 11, 2018 at 14:15
  • 10
    \$\begingroup\$ For future reference: this is called a "snake" draft. \$\endgroup\$ Commented Oct 11, 2018 at 19:44
  • 3
    \$\begingroup\$ Turning your party into competitors sounds like a terrible Idea. Also given if there's only say one 17 to pick from deciding who gets it would create so much resentment. I would say this is probably the worst, anti-cohesive, resentment generating method I've ever seen. This stat picking system is inherently adversarial. \$\endgroup\$ Commented Sep 19, 2021 at 21:08

Tarot Ability Scores

My group successfully used Tarot ability score generation; despite the name, it doesn't actually require a tarot deck. This involves taking a deck of cards (we just used standard playing cards) and using them to simulate dice rolls.

The first step has you preparing your drawing deck. This will consist of 18 cards, all of face values 1-6. The amount of each valued card in the drawing deck is selected by the DM; for example, if they want a low-powered game, there should be fewer 5s and 6s. The array we used (for a D&D 5e game) was 6,6,6,5,5,5,5,4,4,4,4,3,3,3,3,2,2,2; this has a slight bias toward medium scores in the 12-15 range.

Once the drawing deck is prepared, the player draws six cards and lays them out. Next, they draw another card and stack it on top of an existing one (but not more than one new card per stack) until they have drawn six more cards. They then repeat the last step with the last six cards, leaving them with six stacks of three cards.

Total up each stack, and the value is one of your attributes just as if you had rolled it.

This method can be done with cards face-up or face-down. Face-up cards allow the player to calibrate their abilities toward their desired result, while retaining some randomness and organic feel. Face-down cards leave choices out of the player's hands, but does ensure an average ability spread that's more fair than simple rolling.

We found the face-up version to be the best option we have tried yet.

(Method source and additional suggestions here)

  • 3
    \$\begingroup\$ You could do all sorts of variants. First card is face down, the rest are face up, for example. \$\endgroup\$ Commented May 26, 2019 at 18:19

Disclaimer: I am only familiar with D&D 3.x; I hope that this answer remains applicable to a broader set of editions, but cannot guarantee it.

TL;DR: Skip to the latest non-italic section name "Tailored Random Ability Score Generation" to see the proposed method.

Random Ability Score Generation is only partially the problem.

You have noted two different problems.

In other cases, a player may have their characters repeatedly commit suicide-by-monster so they can try to reroll for higher stats, which can be quite frustrating for the GM and other players.

The player is not satisfied by their character itself.

For example, one player character may end up much weaker or much stronger than the rest of the party, which can result in a poor experience for some of the players.

The players' characters do not function well together.

The Ability Score Generation plays a role in each, but is not the sole contribution factor. I still remember a terrible experience playing a sneaky character, where inexplicably the monsters would always immediately locate and attack my hidden character at the beginning of combat, like any other character. I felt cheated by the DM, who in turn considered that it was only fair that my character be targeted. In the end, I switched to another character and play style. My character, in isolation, was not the issue; my DM handling of the character was not to my satisfaction.

This answer will focus on Ability Score Generation, just remember it may only be part of the issue.

Why yet another answer?

There are already many suggestions in this thread of random ability score generation which give the player some agency. I particularly like the Collective Generation from Oblivious Sage which allows choosing from multiple arrays and the Deck of Cards from Elial which allows control during the generation of the array.

I believe they are good, yet I believe they lack a critical component: defining, more precisely, the problem they are attempting to solve, so as to obtain a set of goals by which solutions can be measured.

This answer will therefore endeavor to first establish such goals; and then only propose a solution which satisfies them all.


If your players are anything like me, then they invest a lot of time in creating a character. They'll start from an idea, refine it, research material, refine it further, drop some concepts, add some others, etc... Finally, after hours upon hours of labor, they'll end up with a pretty good idea of their character's strengths and weaknesses, background and future development, etc.

After so many hours spent refining and enriching the character, if they obtain an array of ability which either mechanically does not allow the character or thematically does not fit the character, then of course they'll be extremely disappointed.

Party Balance

A balanced Ability Score Generation, whether random or not, will not however solve party balance issues.

At one extreme, 8/8/8/8/18/8 before racial adjustment is a strong array for a Druid, but utterly crippling for a Monk or Paladin. At another extreme, 16/16/16/16/16/16 is excellent for Monk and Paladin, but is actually worse than the previous array for a Druid.

The point I am trying to make here is that Equality of Opportunities does not imply Equality of Outcomes. Some characters are inherently more dependent on their ability scores than others, and not all characters favor the same distribution (spiky vs flat).

Of course, a good Ability Score Generation is NOT sufficient to achieve a good Party Balance, however please do note that a bad Ability Score Generation can definitely sabotage it.

It is thus in the interest of the Dungeon Master to work with their players to identify the needs of their respective characters and ensure that no player feels left out by whichever Ability Score Generation method is selected.

For example, taking the Deck of Cards proposal, the composition of the deck can favor either extremes (more 6s and 2s) or medians (more 5s, 4s and 3s), with the former benefiting SAD1 characters and the latter benefiting MAD1 ones.

1 SAD: Single Ability Dependent; MAD: Multiple Abilities Dependent.

Tailored Spot Light

Where I make a slight digression about Party Balance and Session 0.

From experience, Party Balance is not as much a matter of capabilities, and more a matter of Spot Light Time. What matters is that in each session, each player should have roughly the same amount of time to "shine", where the scare quotes are used because different player/character pairs will shine differently.

I encourage the DM and players, in the session 0, to establish each character's roles in the party. For example, a Sneak could be:

  • Primary Debuffer/Scout: their main role, where they should outshine anyone else.
  • Secondary Damage Dealer/Party Face: their secondary role, where they can efficiently support others.

It is fine if multiple players share a Primary role, or if a role is left mostly unaddressed: what matters is coordinating expectations.

This also gives information to the DM as to what each player is coming for in the subsequent sessions. If the DM was planning a social game and one player starts explaining they've got this really cool Pyromaniac idea, it also gives time to address the discrepancy before any party has sunk too much time.

And finally, it should help the DM tailor their approach with each player. A player cannot create a cool concept character without the DM's approval and assistance. Approval to ensure that the character fits the narrative and assistance because the character's background and evolution will have to be woven into the narrative.

Ideal Random Ability Score Generation Requirements

Ideally, the Random Ability Score Generation should:

  • Feel random: players picked it for the thrill, let them experience it.
  • Feel fair: no player should feel cheated.
  • Feel empowering: players should feel they have some control over the experience.
  • Be exciting: no player should apprehend the step.

This is, obviously, very subjective, as it is all about the feeling of the players and not about any mathematical outcome.

There are, however, guidelines which can be extracted to inform the process:

  • Feel random: some random process, such as dice or cards, should be included in the method.
  • Feel fair: sharing should avoid envy and jealousy from creeping in.
  • Feel empowering: the players should be driving the process, such as rolling the dice, drawing the cards.
  • Be exciting (tailor): the players should have some degree of agency to tailor the outcome to their particular needs.
  • Be exciting (fast): the players should not have time to get bored.
  • Be exciting (simple): even players with rudimentary mathematical skills should feel at home.

Tailored Random Ability Score Generation

How to randomly generate Ability Scores for Fun and Profit.

Generating Ability Scores is laying the first stone of the campaign to come, it should be an exciting shared moment to kick the campaign off in style. Follow this quick guide to start the party!

  1. Each player creates a base array of 6 numbers, where each number is generated by rolling 3d6 and dropping the lowest. The base arrays are placed at the center of the table.
  2. Each player decides on a base array. It is perfectly acceptable for multiple players to opt for the same base array.
  3. Each player receives a pool containing the numbers [6, 5, 4, 3, 2, 1].
  4. Each player assigns the sum one number from the base array and one number from the pool to each individual Ability Score of their character. No number from the base array or the pool may be used twice.

That's it.


First of all, the proposed method should satisfy all criteria above: the players are driving the process (empowering), rolling dice (random) and sharing the base arrays and pools (fair). The whole thing is accomplished quickly (exciting (fast)) and does not overtax one's mathematics capabilities (exciting (simple)). Furthermore, the players are given sufficient agency to adjust the resulting array to their needs (exciting (tailor)): either uniforming the scores, or skewing them further; choosing to keep a very low ability, or shoring it up.

Secondly, statistically speaking, the method produces relatively random results, yet with a sufficient degree of customization that different needs are catered for.

Using AnyDice, statistics for [highest 2 of 3d6]:

  • Average: 8.46.
  • Median: 9.
  • Minimum (0.46 %): 2.
  • Maximum (7.41 %): 12.

From a player's perspective, considering 6 abilities are generated in this fashion using:

ABILITIES: 6 d [highest 2 of 3d6]
loop P over {1..6} {
    output (P @ ABILITIES + (7 - P)) named "Ability [P]"

See output here.

This works well for a caster (SAD):

  • 36.98 % of achieving 18 in the top score.
  • 73.60 % of achieving at least 17 in the top score.
  • 92.90 % of achieving at least 16 in the top score.

This also works well for a Barbarian (Str > Con):

  • 6.73 % of achieving 17 in the second top score.
  • 34.24 % of achieving 16 in the second top score.
  • 69.29 % of achieving 15 in the second top score.

This also works well for a Monk (Str > Dex > Con > Wis) or Paladin (Str > Con > Wis > Cha):

  • 12.46 % of having 4 10+ in the base array: 13+ in fourth top score.
  • 38.81 % of having 4 9+ in the base array: 12+ in fourth top score.
  • 70.75 % of having 4 8+ in the base array: 11+ in fourth top score.

Note: I am not quite sure how to compute the chances of a 13+ or 14+ in the fourth top score assuming the player assign +3 to the top, +4 to the second, +5 to the third and +6 to the fourth (aiming for uniformity). I'd appreciate help from an AnyDice guru.


Thanks to Oblivious Sage for their Collective Generation method; their answer was the first time I encountered the concept, and I believe sharing the results of the dice rolls is an excellent way to start a shared story.


Consider placing high/low caps on the total stats rolled.

One method that you can try is setting a maximum and/or minimum value for the stat total rolled by the players, and have them reroll if their total is above the maximum or below the minimum. This allows some of the variation and randomness that dice rolling offers while preventing too much disparity between party members or abuse by players.

For example, if you decide to set the minimum cap at 70 and the maximum cap at 80, one player might roll: 18 6 11 9 14 16, for a total of 74. Another might roll 14 16 13 12 10 14, for a total of 79. If your third player rolled exceptionally well: 18 16 15 11 13 12 (total 85), and your fourth player rolled poorly: 10 13 9 11 14 10 (total 67), both would need to reroll until they landed a total between 70 and 80.

You might need to experiment a little to find the right limits for your group. You could increase the difference between your caps if you wanted to allow greater disparity between players, only set a minimum cap if you're only worried that some characters will be too weak, or only set a maximum cap if you're only concerned with players killing their characters for a chance to reroll a better one.


Dice pools for stats

(This one is totally unique here, as far as I see.)

Rather than having players assign points to stats, give each player a pool of dice that they can divide up between all six stats. Then they roll each stat's pool and whatever they get is the value they have for that stat. I found that this works best with the parameters being:

  1. 24d6 as the pool to divide (this works out to an average of 4d6 per stat)
  2. Minimum two dice assigned to a stat (which should be rolled as 2d6+1), no maximum.
  3. After each pool is rolled, take the highest 3d6 (or the highest 4d6 if that isn't over 18, otherwise reroll if you want to do it that way), and that's that character's stat.

This gives some of the customization of point-buy, allowing players to choose which stats they care about and which they're willing to let go. However, it's still random enough to generate the kinds of interesting characters that rolling is famous for. In my experience, it tends to generate characters with a 16+ in the stats they really care about, but with at least one surprisingly poor stat, and often one that's unexpectedly good. And by all coming out of one pool, it ensures that there will be around the same number of good stats vs bad stats in the party, without mandating it.

Use filtered group-wide arrays

(This is very similar to some other answers, but this variant worked very successfully for me.)

Have everyone in the group roll up one (or two, depending on number of players) sets of 6 stats, using whatever dice rolling method you prefer (3d6, 4d6-drop-lowest, 4d6-cap-18, etc.). Calculate the total modifier for each array. Then, throw away the highest set and the lowest set (or the two highest and two lowest), and let the players choose between the remaining. Then, they can assign the stats as they choose.

This has the advantage of pushing the available arrays towards the median distribution, so there isn't one "obviously best" array that everyone takes.

Generate a lot of stats and let players pick

(Again, this is somewhat similar to other answers, but I've used a variant successfully)

Another option is to have everyone roll up three sets of stats in order (no swapping stats around). Then, put all the triplets together and let each player choose one triplet to claim, from which they can use one stat array. This should generate enough rolls that no triplet will be a total disaster, and lets players find a set of stats that matches the character they want - all without making it so loose that it's equivalent to rolling until you get the right arrays.

  • \$\begingroup\$ For the "triplets" in order - does this mean that you can choose one of the 3 values for each stat? eg if the 3 sets are labelled a,b,c: STR a, DEX b, CON c, INT b, WIS c, CHA a? \$\endgroup\$
    – Adeptus
    Commented Feb 8, 2023 at 23:16
  • 1
    \$\begingroup\$ @Adeptus It's been a while, but I think I was referring just to the set of three stat arrays. So with four players, you'd get {a1,a2,a3, b1,b2,b3, c1,c2,c3, d1,d2,d3}, and then one player chooses B and then chooses between b1, b2, or b3, the next picks C, etc. It's basically the same thing as just rolling up (player * 3) stat arrays, but because some choices block other choices (if someone wants to use b2, no one can use b1 or b3), it encourages discussion and compromise. Maybe b2 is perfect, but a3 would also work, and b1 is the only good choice for someone else, so you take a3 instead. \$\endgroup\$
    – Bobson
    Commented Feb 9, 2023 at 17:21

One thing I always suggest is that everyone try to put aside their preconceived notions about the importance of ability scores for a good amount of time and just see what happens. Depending on your play-style and the edition, ability scores may have a much smaller effect in practice.

To give a few examples:

Generally, the older the edition, the less rules you’ll find that specifically reference ability scores, and the modifiers tend to be smaller.

How often the DM calls for rolls and what difficulty factors they choose will also change how much ability scores affect the game.

Some players play in a style that tries to minimize the number of rolls they have to make and seek circumstances to maximize the odds when they do roll. They always look for creative and low risk solutions. This tends to minimize the effect of ability scores.

But I would encourage you to simply try it and see how it works with your edition and your group rather than analyze such factors.

Also, it can help to try to think more in terms of the group than the individual. Yes, sometimes one PC is consistently more effective than the others, but that doesn’t mean that the others were ineffective or that they didn’t contribute in important ways. Would that PC alone have fared as well without their allies?

If you have given it a fair shake, and if it isn’t working for your group, then the other answers provide good solutions.

  • \$\begingroup\$ I agree with this, but it would be nice to have more specific examples since this is aiming to be a canonical question... Which editions are more/less dependent on ability scores, and what changes can you make to your play style to be less dependent on ability scores? \$\endgroup\$ Commented Oct 11, 2018 at 13:43

Don't roll for effectiveness. Roll for distribution.

Greg Stolze's One Roll Engine game Reign takes this rough approach to random character creation. You pitch a bunch of d10s for character creation and each set of different numbers gives you benefits flavored after the number you rolled that get more powerful as you get more numbers in the set. But each individual die has the same total contribution to character creation - that is, if you roll a whole bunch of singletons and somebody else rolls two matched sets, you'll have a lot of skill bonuses spread out widely, and some extra starting gear or treasure, and they'll have a few good core stats, but not a lot of skills outside of them. In the end you'll have "spent the same number of points" at character creation.

So how do you adapt that philosophy to D&D?

Split 25.

Roll 4d6 drop lowest three times. If any of them came up less than 7, treat it as a 7. Subtract each of those three numbers from 25. There's your stat array.

So if you roll 15, 14, and 13, then you also have a 10, an 11, and a 12. If you roll 18, 17, and 16, you also have a 7, an 8, and a 9. When you get a good roll, you also get a bad roll. When you roll average, you get another average roll. You'll always have three odds and three evens.

Ultimately this isn't going to be perfect. Some characters are better with unbalanced stats than others, usually casters who can use magic to make up for physical deficiencies. But it works at making the rolls as a whole feel less unfair.


As DM, you choose what matters.

There are two sides to avoiding problems related to the utility of a character's ability score: the player/character side and the DM side.

On the character side, the most straight-forward approach is to use a point-buy system for abilities and that has been an option in the rule books since 3rd edition. The method of rolling 6 sets of 3d6 and choosing one, or of rolling 3d6 twice and selecting one probably have the strongest effect, of those offered in the rules, for avoiding a "dismal" set of scores while not likely creating an overpowered set and still allowing for random generation.

More powerful than the character side of avoiding this problem is the DM's side, however. As DM, you have control over how much impact high and low scores have. If toe-to-toe combat of rolling d20 and adding ability bonuses is not going to produce desirable results in your game, don't do it. The same goes for skill/ability checks. Don't call on the whole table for a listen check, for example, instead, while all the muscle-bound and brainy types were sharpening blades and reading spellbooks around the campfire, the wimpy character was collecting firewood at the edge of camp so he's the one who heard the approaching orcs. Let him get an extra shot off at the beginning of the combat.

Create situations that reward the player with the crappy character by playing to what they're good at: solving puzzles, social interaction, or whatever. Then judge the results by what the player says the character does or says instead of an intelligence check or diplomacy roll.

Two final points: 1. This is a cooperative game, not a competitive one, and players and their characters should be working as a team. When they don't the game world should punish them. 2. If you let your players be kind of their characters' guardian angels, you can make playing the game more about their fun, and less about their game pieces.


Set minimums on the number of well rolled stats and maximums on the number of poorly rolled stats

No one likes playing a poorly stated PC. It takes away from the fun of the game. One of the primary draws of D&D is that the players will be playing characters who are exceptional (if the characters aren't exceptional why isn't everyone adventuring?!).

To achieve these aims, while also allowing for some randomness it can be useful to set minimums and maximums for set of rolled stats.

The method is use when my players choose to roll:

  • Rolling happens while I'm present
  • Players roll 4d6 and take the highest three. They do this 6 times to generate a stat array.
  • Out of the 6 rolls at least two must be 14+ (before any modifiers are applied)
  • Out of the 6 rolls no more than two of the rolls should be less than 10 (before any modifiers are applied)

Once a player has rolled a set of stats that satisfy all of these criteria they can assign their array to the abilities in any order (using a roll no more than once).

You could take this one step further and make the stats be rolled in order (Str, Dex, Con, Int, Wis, Cha


Have each player roll (a fixed number of) multiple characters and choose one to play, one (or more) as backups, and turn the rest over to the GM for use as NPCs.

This is the default character generation method for Adventurer, Conqueror, King (aka ACKS), a recent D&D B/X clone. Players roll five characters, play one with a second as a backup, and the remaining three become NPCs.

Note that, because you're choosing from complete sets of stats, this method works with "rolled in order" character generation, unlike some other methods which implicitly require "arrange to suit" assignment of stat values.


Roll 5

One of my friends came up with this method, which I find quite entertaining. Only roll 5 times. Add up the modifiers of those 5 rolls, and then your last stat is determined so that everyone has the same total modifier.

So for a moderately powered campaign, the DM might say your total modifiers should add up to +8. If your first five rolls are really great, your sixth stat will be pretty terrible, and conversely if you roll terrible for your first five rolls, your six stat will be awesome. As a DM you might want to disallow really extreme sixth stats... Then again you may not want to as long as your player(s) are okay with it. Anecdotal but I played a Con 1 bard in one campaign and it was great ... had the campaign been a bit more combat/dungeon focused and less urban-intrigue it probably wouldn't have worked so you'll definitely want to consider how much to allow.


A Method To Use For Rolling Dice

My nephew (D&D 5e campaign) had us roll up characters using 4d6 drop 1 (the default method in the PHB) arranged to fit abilities as desired. His boundaries were:

  1. "If your total ability bonus score total is +10 or greater, either re-roll or modify a roll down to get to +10"
  2. "If your total ability bonus score is +3 or less, re-roll"
  3. "If you do not have at least one score of 16 (or higher) after rolling, you may roll again if you wish, providing 1 and 2 are complied with, but you are not required to."

    \$ \begin{array}{|c|l|} \hline \text{Score} & \text{Modifier} \\ \hline 2\text{–}3 & −4 \\ 4\text{–}5 & −3 \\ 6\text{–}7 & −2 \\ 8\text{–}9 & −1 \\ 10\text{–}11 & +0 \\ 12\text{–}13 & +1 \\ 14\text{–}15 & +2 \\ 16\text{–}17 & +3 \\ 18\text{–}19 & +4 \\ \hline \end{array} \$

I ended up with +12, due to a 16, 16, 15, 12, 13, 14 (I was HOT!) I complied with Rule 1 and dropped the second 16 to a 15, and the 14 to a 13. That left me with a +10. (DM okayed this).

My brother (his dad) was ice cold. He barely got the +4, and had one score of 16. He kept his scores. We played and had fun. No worries.

With this method, you can set the +bonus range to anything you'd like, and perhaps make it narrower than what my nephew allowed. (such as +8 to +4, or whatever).

What does this do?

It keeps the randomness of die rolling in the mix, but it also truncates the results to keep the variation from going too far afield.

About point buy

If using the 27-point buy (page 8. Basic Rules, D&D 5e) one can arrive at +6 or +3 (before racial adjustments) in a few different ways (and a variety of points in between).

  • Buy 15, 15, 15, 8, 8, 8 for 3x (+2) and 3x (-1): aggregate +3.

  • Buy 13, 13, 13, 12, 12, 12, for aggregate +6,

  • Buy 14, 14, 14, 10, 10, 10, for aggregate +6.

And just because we can ...

  • Buy 14, 12, 12, 12, 12, 12, for aggregate +7 (@ShadowRanger, thank you!)
  • \$\begingroup\$ On point buy: Unless I'm forgetting some point buy restriction, +7 is possible, if mostly useless (maybe usable for jack of all trades who want to be decent at everything given bounded accuracy, but suboptimal for class abilities and most stuff in combat): 14, 12, 12, 12, 12, 12. \$\endgroup\$ Commented Nov 4, 2019 at 21:53
  • \$\begingroup\$ @ShadowRanger Right you are, I'll add that in as an example. \$\endgroup\$ Commented Nov 4, 2019 at 22:13

What I have tried with great success was simple:

Shared rolls

Each of 6 players rolled once, standard 4d6 drop lowest. That was their stat array. Everyone was free to assign them as they wish, and racial bonuses made characters more varied than it looks like.

If there was less than 6 players, then depending on the mood and difficulty of campaign I simply gave one 16 / 17 / 18 to the pool. Or one 8. Or both.

This worked best when I was DMing for teenagers. I wasn't much older either. It saved quite a lot of drama about 2 points total difference I had to deal with.

Other thing I tried, the way we play now is:

High and low stat given by DM

In the campaign I'm DMing, and DMed before, each character has one 18 and one 8 before modifiers to begin with. other stats are rolled in a way1 that can't generate 18, and can't generate lower than 8. That way each character is guaranteed to have a stat that shines and a stat that's a weak spot.

It is especially fun when players are playing weak spots. Imagine mage tower, and rogue with 18 dex and 8 wis...

This worked like a charm when playing with adults that just want their characters to be good at something, don't mind playing flaws, and are not envious about minor differences in points total.

1 that's campaign specific and probably copyrighted


Allow limited rerolls

Sometimes players roll something unplayable. I had a guy roll something between 10 and 13 across the board. I told him to reroll the entire thing.

I had another player roll 3 stats at 15 and above and 3 others below 7. While this is playable, it wouldn't be very fun, so I allowed him to reroll his two lowest. He rolled average on the rerolls and decided to play a barbarian with 3 int (-2 from racial penalty).

I had somebody roll a slightly above average array: 18, 14, 14, 14, 7, 5 (or thereabouts). He wanted to play a 3.x Paladin, and while that's almost a good 3.x Paladin array having 2 stats at -2 before racial penalties is not viable. I denied a reroll on this because that was a very playable array, even if it wasn't Paladin viable. If a player's best 3 ability scores are decent and they didn't roll too badly on the other 3, I rarely permit rerolls.

Aside from that, there's no hard and fast rule. I consider the type of character a player wants to make, but if they roll a good or decent array for a SAD class (Single Ability Dependent, ex. Wizard) I almost never allow it. Sometimes I allow the player to roll off a low stat, sometimes an average stat, and sometimes a mix of high, mid or low. I rarely need to offer rerolls at all.

Rerolls are a privilege, not a right, and my players know this. Everything about the process is subject to GM fiat based on whatever I feel like I need to do to let my players have a chance at having fun and being successful in-game. If they want to play a heavily MAD class (Multiple Ability Dependent, ex. Paladin in D&D3.5) and don't get the stats for it, no dice. After all, it's not like you'll have a chance at running a decent Paladin or Monk in point buy.

  • \$\begingroup\$ The problem with this method as described is that it seems very subjective. It is hard to implement your suggestion - essentially it is "look at the stats and make a decision", with a couple of example decisions that you made. \$\endgroup\$ Commented Oct 11, 2018 at 14:58
  • \$\begingroup\$ Neil - It is subjective, but not impossible to implement. The GM and the players need to have a good enough understanding of the game to know what they can do with certain arrays and what arrays are hard to work with in general. You can't make a 1-size fits-all approach with this, so it's better to leave it up to GM discretion. If the players think that it's an unworkable array, they'll let you know. \$\endgroup\$
    – VHS
    Commented Oct 11, 2018 at 18:43

At some point you need to ask yourself why you're rolling for stats. Do the players want the chance of having dramatically good or bad characters? Do they just want a chance of getting a good or bad stat somewhere they wouldn't normally put it to mix things up? I've seen people using rolled stats but then layering on so many mins and maxes and rerolls and swaps and calculations that it removes most of the randomness and adds a lot of effort.

If you have the kind of players who are overly concerned with balancing (or optimising) stats a point buy is probably going to be the least likely route to cause problems.

That said here are some suggestions I haven't seen here yet:

Change the dice you use to roll:

Instead of 3D6 try different combinations of dice that either narrow the range to keep the minimum and maximum closer together or create a steeper bell curve to make really high or really low rolls less likely.

Here are some examples: https://anydice.com/program/11d08

Random point buy:

Use a point buy system but find a way to randomly distribute the points so you have randomness in what is high and what is low but everyone is still roughly equal. Not sure if there's a good way to do that though without having to use something like http://stuff.nathan.net.nz/dnd


The Random Elite system

Goal: to give all players roughly equivalent Attribute scores (total plus value), without removing the fun and randomness of rolling dice.

My solution is the Random Elite.

Each player starts with an Elite Array/Grid. For example 13, 11, 11, 9, 9, 7. These values are assigned to the six attributes like a normal Elite Array system.

Once the initial values are assigned, each player rolls a certain number of d6. Each die result adds 2 to a given attribute. 1 - STR 2 - INT 3 - DEX 4 - WIS 5 - CON 6 - CHA

Optionally, if the roll would result in an attribute being higher than a certain value (17 or 19, usually), the player must a assign that bonus to a different attribute of their choice (so long as that would not bring the second attribute over the limit).

So for example, Harry wants to play a Barbarian with might thews. His GM Will is playing a semi-low power level campaign using the array above, plus five bonuses. Harry assigns his array, giving him starting values of:
STR - 13
DEX - 11
CON - 11
INT - 7
WIS - 9
CHA - 9
Harry then rolls his 5d6. He gets 1, 2, 6, 2, and 3. Referring to the chart he adds +2 bonus to the indicated attributes, giving him a result of:
STR - 15
DEX - 13
CON - 11
INT - 11
WIS - 9
CHA - 11
Not exactly the dumb, ugly, mighty Barbarian he was hoping for but gives him the option of hidden depths.

The Random Elite system can be adapted to the level of power and randomness you and you players desire. Weaker grid with more dice (all 9s for starting values plus 14d6), stronger grid with fewer dice (15, 15, 13, 11, 11, 9 with only 2 dice), or even a strong grid with more dice for that superhero power level campaign. My general preferred level when I DM is 13, 13, 11, 11, 9, 7 + 7d6 (redistribute over 19).


Try the character first

I see this more often as a fear than a reality. And it only becomes a hinderance is you let it be one. Dnd is not a video game stats don't matter much unless you want them too.

Focus more on roleplay and class features.

If one player having high stats makes other character useless, then you may be putting too much emphasis on stats. Low states should not hinder the spotlight.

First high states does not alter access to skills or abilities. No matter how high the warrior stats he still can't cast fly or cure wounds and he probably doesn't know anything about stealth or acting.

A character is defined by their low stats not their high ones.

Clever players can use low stats. In a recent game I played I rolled very high, (16,18,15,13,17,6) so I decided to play a subclass everyone says is underpowered, a monk of the four elements, other players had more normal scores but had more "optimized" subclasses. We all had a ball, and high stats did not stop my character from getting downed all the time, (she was the closest think to a tank we had) she was a lizardfolk shaman who could not read, barely spoke common, and was perfectly happy to cook and eat humanoids (just not reptilian ones). One bad score became far more character defining than all the good ones. She also tried diplomacy all the time to the utter horror of other characters. Other players had a ball, because I did talk about the character first. Other characters would often "jump on the grenade first" to prevent her trying diplomacy. She once offered to cook a mourning feast out of a dead child for the family and threatened to bite the faces off annoyances all the time. She was smart but from a completely alien culture, think Crocodile Dundee.

Low scores can be FUN

I also played a dwarven wizard, whos highest score was a 14, (12,7,12,14,6,7) again had a ball. Played up the fact he was a wizard form a race naturally resistant to magic (back in older editions dwarves had resistance to magic) he spent all his time trying to get better at magic, but kept shorting out his own magic. He was absent minded, talked in technobabble (arcanobabble?) and needed glasses to see. I leaned into it and we gave him the wild magic feature. He rode around on Bulette that started life a mule. He killed a dragon with an accidental wall of force, changed the race of two of his party members (and turned another one purple), became immune to force spells, and set 3 towns on fire. He was destruction on wheels and often blissfully unaware of the chaos left in his wake.

A bad character can actually be a great thing to play, the farmer who lost their home and had to turn to adventuring, the exiled noble, the kid. there great ways to twist low scores into opportunities, think about why your character might have a low score. lean into it, the person who thinks they are a great negotiator when they are not, the barbarian that is dumb as a box of rocks, the "thief" that can't pick a lock, the wizard who who is perpetually ill, all defined by low stats and wonderful to play. I often encourage players to add major flaws to a character for this reason. If I don't have a character concept for a game I always roll and let the roll inspire the character.

Advice from older editions.

Older edition in which straight rolling was more common often explicitly offered advice to not make players play characters they did not want to play but encourage them to try. 2nd edition offers some great advice for this. Common reason was the unplayable character worry, as an old DM put it "I'm not going to make you play a character that is clearly unfit to leave his house" if you don't roll a single score above a 5 just let them reroll. But the take away is to try out the character first, characters are defined by flaws more than anything else and unique characters can be way more fun to play than generic fighter #7.

5th edition is great for this simply because stats are not that important in the game. the impact of stats is quite small and swamped by other features like skill selection, backstory, and class features.


Inspired by @DavidCoffron 's answer:

Ranked - Choice rolling

There are three parts, and you can do the first two in either order depending on how much foreknowledge you want the players to have.

Part A: Rank Stats

Everyone ranks the 6 stats in order of importance to themselves.

Part B: Roll stats

Everyone rolls for the six stats in order. All of the rolls for strength are pooled, and ordered. The same for the other five stats.

Part C: Apportion Stats

Everyone who's first pick which was not tied gets that stat and removes it from the pool.

If there are ties for first pick, check 2nd picks to see if those are tied with anyone, if not everyone get

Once everyone has one stat, cross that choice out of their ranking and renumber the choices so they are 1-5 and then repeat this process. Once you've gone through this 5 times everyone gets their last stat from the remaining choice in it's pool.


      ST  DX  CN  IN  WS  CA
Anne  12  17  11  12  8   13
Bob   9   14  16  12  13  13
Chris 13  9   12  15  12  9 

Anne wants to play a fighter and so chooses ST,CN,DX,WS,IN,CA Bob wants to play a cleric and so chooses WS,CN,ST,CA,DX,IN Chris wants to play a rogue and so chooses DX,IN,CA,CN,WS,ST

Everyone gets the best stat for their first choice:

      ST  DX  CN  IN  WS  CA
Anne  13                      
Bob                   13      
Chris     17                 

Anne & Bob are tied for their second choice so Chris gets his 2nd and we look at A&B's 3rd. Bob gets the #2 St, and Anne gets the #2 DX.

      ST  DX  CN  IN  WS  CA
Anne  13  14                  
Bob   12              13      
Chris     17      15           

Anne's remaining choices: CN,WS,IN,CA Bob's remaining choices: CN,CA,DX,IN Chris's remaining choices: CA,CN,WS,ST

A&B are still tied for for CN so we put their (now) 2nd choice first. This makes B&C tie, and we end up with Anne getting WS, Bob DX and Chris getting ST

      ST  DX  CN  IN  WS  CA
Anne  13  14  16  12  12   9  
Bob   12   9  12  12  13  13  
Chris  9  17  11  15   8  13   

The above is the final result, however if the rolls are made before ranking choices people will likely different choices as to which stats to prioritize... Also the example rolls were actually not very great.

I'm not sure this a great idea, but I'm going to post it anyways ... I can always delete it later if I've overlooked something.

  • 1
    \$\begingroup\$ I'm not sure I like this system, but it might represent an interesting compromise between other collective/drafting ability score systems and the really old-school "3d6 in order, pick a class based on what you got". If nobody rolled a good strength then by golly the party is going to be full of weaklings. \$\endgroup\$
    – Oblivious Sage
    Commented Jan 12, 2021 at 22:31
  • 1
    \$\begingroup\$ If you have not used this in play, but are instead suggesting it as a solution baed on an untested idea, that makes for a much less useful answer (or opinion) than if you have used this with a play group and found it to be successful . \$\endgroup\$ Commented Jan 12, 2021 at 22:46

Choose class first, then roll

  1. Choose the character class first, then roll 4d6-drop-lowest down the line.
  2. If the prime requisite is lower than 13 (roughly a 50% chance), you may choose to reroll the entire array (again and again, if necessary).

This ensures that the character either is competent in his area of expertise or has surprising strengths in other areas (i.e. you'll get a lot of fighters with Str 13 and a few with Str 9 and Charisma 18).

This method allows the players to choose character class strategically ("we need more fighters for the front line") and also occasionally leads to characters you almost never see with other character generation methods.

  • 4
    \$\begingroup\$ How does this prevent characters with a single 13 and a bunch of rubbish? You say"either / or" but I'm not sure how it'd stop this. \$\endgroup\$
    – Erik
    Commented Apr 8, 2021 at 21:06
  • 1
    \$\begingroup\$ Hi Johann Mitland! Welcome to the RPGSE! /wave :) \$\endgroup\$
    – Rykara
    Commented Apr 8, 2021 at 21:43
  • 1
    \$\begingroup\$ Welcome to our stack! Please take the tour to learn more about how we operate and you can also visit the help center for even more information. We are different than forums in that just submitting your idea isn't enough. The statements that you say do need to be supported - in this case by either your own experience or in something you've seen. The notes that Erik has above is a clear issue with your answer and do need to be addressed. \$\endgroup\$
    – NotArch
    Commented Apr 8, 2021 at 21:51

The Heroic method of rolling characters calls for 2d6+6. This prevents scores under 8 but actually increases the odds of an 18.

An alternative would be 2d4+6. Max score of 14, min score of 8, thus allowing some randomness but still keeping everyone in the same range.

  • 1
    \$\begingroup\$ +1 I like this method as it keeps things simple. I might implement 2x 2d6+6 and 4x 2d4+6 so that there is less chance of OP munchkin nonsense. :) It will also help those who wish to multi-class which can be tricky in RAW if you roll a set of dire scores. \$\endgroup\$
    – Senmurv
    Commented Apr 21 at 9:25

Have players roll an ability score array for the entire party

I've used this method as both a player and a DM, with great success:

  1. The party collectively rolls stat values for 1 plus the number of stats (e.g. for D&D 5E, you'd roll 7 sets of (4d6 drop lowest)). These can be divided up however you wish - with a party of 6, you might have each player and the DM each roll one stat value.
  2. The party and the DM collectively decide which score to drop from the array.
  3. Each player decides how to allocate the remainder of the stat array for their individual character.

This method allows for the unpredictability of random rolls, while also keeping the characters balanced by having them use the same array. Additionally, the fact that the entire party decides which value to drop allows them to decide as a group how powerful they want the party to be, within the bounds set by the rolled stats.


There are quite a lot of methods to not have random results or limit the randomness of results of 'rolling ability scores'. However, applying those methods doesn't particularly mitigate any of the problems of rolling scores - it simply makes you roll them less, or not really roll them at all. (There are quite a lot of clever ways to basically remove the randomness from the process while seeming like it's still there).

This is actually a spotlight time/screen time question.

Let's set aside that classes or feats or whatnot can have a far greater effect on character mechanical power than ability scores in many games (or editions of DnD). Someone with greater numbers on their sheet than other players will succeed more often than other players at in-game tasks, barring GM intervention. This will lead to them being called on to complete more tasks (as players and characters will naturally want to succeed more often), and gain more approbation for success. Humans are wired this way.

Some people will find that unfair or resent their lack of screen time/success in comparison to others, especially if it is determined by a single set of die rolls at game start. There are quite a lot of ways to give people more screen time, or help people to enjoy the unfolding of the story or other aspects of the game without feeling that they either need as much or more screen time/success/whatever as other people.

This can take the form of the GM artificially granting screen time to some characters or making tasks those characters perform less mechanically demanding, stronger characters having reasons to value or call on weaker ones (whether roleplaying or mechanically) and thus give them screen time and status, altering the mechanical stats of characters to better reflect the party dynamic, so on. There's reams of advice on the internet about this in many places.

However the first step to incorporating any of this into a situation where rolled ability scores cause a mechanical power discrepancy amongst characters is to recognize the nature of the problem. This is about screen time, and it is not fundamentally different to a situation where you have a player who gets a lot of screen time due to social skills and personality and another that has weaker social skills and is less outgoing, leading to feeling like they are not contributing to the game.

You can solve both by largely the same methods.


OK, well I think the real reason anyone wants to roll ability scores is so that they can start with at least one high stat. In my games, I use point buy, however, the first "15" that you buy, is an 18 instead. It makes for a slightly more powerful party, but everyone feels more heroic that way. And it also maintains balance.


I just have them keep rerolling bad scores until everyone's about even. It ain't rocket science, just keep it simple.

For example:

I tell Josh to reroll the 8 or Bruce to roll a d6 and put it wherever he wants up to 18. If Tom rolls especially good then he just has the best stats but we can get everyone else in the ballpark with a few rolls if it's important.

  • 1
    \$\begingroup\$ Comments are not for extended discussion; this conversation has been moved to chat. \$\endgroup\$ Commented Oct 17, 2018 at 16:07
  • 2
    \$\begingroup\$ Robert, it would be worth detailing how this approach, that you advocate, works at your table. Specifically, how your players feel about it, what sort of impacts you get in your game, and also how you define or assess 'about even'. \$\endgroup\$ Commented Oct 17, 2018 at 16:12

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .