In DnD 5th edition, all classes seemingly have the option of rolling hit points -or- just increasing their hit points by a set value.

That set value is defined as the dice roll's average, rounded up.

Given that it is rounded up (and thus will statistically offer better HP than rolling), why would anyone ever roll for hit points? Rolling seems to provide the worst of both worlds, giving you lower HP than defaulting as well as possibly screwing your barbarian over with a roll of 1, and just risking plain bad rolls in general.

It makes me wonder if I've overlooked any rules that otherwise balances the options. Am I?


19 Answers 19


Because it's "more fun".

To many roleplayers the fun is on the dice. They enjoy every roll, and they like to take risks.

If another game's example is allowed, in MERP hitpoints were also rolled (for some reason it was the only rolled skill). Everytime I leveled up, I had incredible luck. While the average animist had 33 hp, mine had over 50. So no, rolling is not always worst. It is a suboptimal choice, but can provide better results.

Also, I think this option is provided as something like backward compatibility. Old habits die hard on roleplayers, and many would feel a fixed raise spoils the game, while others may find annoying to randomize so much his character's power.

Possibly, if D&D 5 was a new game instead of a new edition of an ancestral game, they would never have provided the roll option.

  • History
  • It's a Gamble
  • It's presented as the default.

The first of these points might be the strongest of them all. Older editions had you always roll your hit points which means that people who are used to old editions may feel a certain nostalgia to roll them rather than take a more optimal average. Nothing wrong with this, it just carries more risk than taking the average (and less you should end up with slightly fewer HP, about 10 over the course of 20 levels, though that's probably not enough iterations for it to really show up that starkly, there's a 16% on a d8 --better on higher HD, worse on lower== chance you'll actually roll better over the course of 19 levels).

There are reasons that Las Vegas, Atlantic City and now casinos all over the US stay open, and make money (recent AC closings aside). People like to gamble. They don't care that they lose eventually, they are willing to sacrifice certain long term loss for potential short term gain. Considering that a PC's life is essentially a short term game. There is a possibility (again a bit under 16%), that you'll actually roll better than the high average with a d8 HD. This chance actually gets better with a larger HD (more numbers on the die over the high average), and worse on a lower HD. Here are the numbers for d6-d12 (to get the %, compare the "At Least" number to the high avg so 19*(4,5,6,7)+1):

  d6   9%
  d8   16%
  d10  21%
  d12  25%

So those odds are what you are banking on when you roll HP. If you play the lotto, those look like pretty good odds. If you like blackjack, not so much.

The last point is also quite significant. Rolling HP is presented as the default, with using the high average a suggested alternative. When you present something as the default, that's what people are going to use.

Looking at this, there are reasons to roll HP, nostalgia, gambling and text presentation are all strong reasons. If your players want to roll, I would make sure that they understand the risk (show them the table). Also ask them about their reasons. Ultimately though, what I do at my table is ask. Do you want to roll your HP or do you want to take the high average? Some players choose to roll and gamble, others choose to take the average. It's up to them, and it can be the difference of a lot of HP, but on average this is 10 HP over the life of an adventurer, that's not enough to take this mechanic out if it's fun for folks.


Rolling dice is part of the game, but rolling dice only once is also very unbalancing. When you miss an attack roll, you can try again next round. When you roll low for attributes, your character will have to live with them for the rest of his virtual life. And if you take in account that for everyone that rolls good, there is one that rolls bad, you create a huge gap in power level, even between characters that should be perfectly balanced (rare enough).

Through the editions of D&D, many of the dice rolls that only happen once in a characters lifetime (like attributes and HP) have been converted to deterministic methods. This way, characters don't vary too much in power. The old methods have been kept for people who like rolling better.

Apart from power levels, the deterministic method also allow you to play the character you want to play. This is about fantasy, so people should be allowed to play what they want. Obviously it has to be balanced, but there is no reason to have someone who's fantasy is to play a slender, dextrous elf play an clumsy, fat elf just because they missed a single diceroll two weeks ago.

I cannot tell you what is good for your group, but for the groups I was part in, deterministic methods have worked out way better in every game we played, not just D&D. Because for feeling cool having rolled good, someone else will have rolled bad and feel frustrated. And having a few virtual power points on a character sheet is not worth having someone in the group who is frustrated. This game is about having a good time for all in the group.

  • 4
    \$\begingroup\$ My thoughts exactly, though not, strictly speaking, an answer to the question posed. That doesn't mean I don't agree whole-heartedly. :) \$\endgroup\$ – TheMorten Sep 21 '14 at 11:43
  • \$\begingroup\$ At the opposite (and maybe a reason for rolling to them) I know players who believe that being allowed to play what you want is just a power fantasy. "How can you have a 20-level build when you don't know which opportunities for prestige classes you will encounter in the game? How can you plan where your hearth will go in advance?" Not knowing if your HD will be good or bad is just another way of having to work with what life gives you. \$\endgroup\$ – Zachiel Sep 21 '15 at 18:50
  • 6
    \$\begingroup\$ @Zachiel A lot of people I know play fantasy games, because they don't want to have to work with what life gives you, if only for a single evening. I belong to that group :) \$\endgroup\$ – nvoigt Sep 21 '15 at 19:26
  • \$\begingroup\$ Voting to delete as not an answer. The question is why would someone ever roll for HP. This 'answer' is about why one would not, which is already presupposed in the OP question. \$\endgroup\$ – Kirt Nov 26 '20 at 20:21

A mechanical reason to roll the dice is that the variability of hit points when only using average hit points is zero, while there is nonzero variability when using dice.

Variability is important for a few reasons.

First, the more variability, the more surprise and uncertainty. Some folks hate uncertainty. Some folks hate a world without surprises.

Second, without variability there are no extremes. It is fun to play non-average characters. A character with exceptionally high hit points is a tank... gets to take risks that others don't. On the other hand, a character with exceptionally low hit points presents a challenge: the character must be more risk averse to survive, and that presents interesting challenges both for role play, and for mechanical play. Some people hate difference. Some people's favorite flavor is the one they haven't tried yet.


Assuming you're playing a game where character death leads to rerolling (not ressurecting).

If you roll HP, and you roll low, you weaken your character, to the point where they are more likely to die. When they do, you reroll a new character.

If you roll high hp, it makes your character more likely to survive.

So if you roll HP for characters, natural selection means that you will spend more time with high HP than low HP.

In this way, characters who roll for HP get weaker, but players who roll for HP get stronger.


A house rule to even the odds

In your question, you say "It makes me wonder if I've overlooked any rules that otherwise balances the options. Am I?"

There is not a rule that you are overlooking however there is a house rule that I use which does balance the options:

The rule is that if the player rolls a 1 on their Hit Dice then they get to re-roll if they roll a 1.

This has 2 effects:

  • Firstly it increases the average roll on the Hit Dice by 0.5, so the average for people rolling is the same as the fixed amount you can choose to take. Thus players aren't punished for choosing to roll rather than taking the fixed amount.

  • Secondly, this takes away the demoralising element of choosing to roll for health and seeing the 1 turn up on your dice. Because that sucks!

A note on probabilities

If you always re-roll 1's on a dice, then you can treat the dice as having 1 fewer side with numbering starting at 2.

Thus for:

  • d6 re-rolling 1s - {2,3,4,5,6} so average roll is 4
  • d8 re-rolling 1s - {2,3,4,5,6,7,8} so average roll is 5
  • d10 re-rolling 1s - {2,3,4,5,6,7,8,9,10} so average roll is 6
  • d12 re-rolling 1s - {2,3,4,5,6,7,8,9,10,11,12} so average roll is 7
  • 2
    \$\begingroup\$ This is a nice house rule which I actually like a lot and will probably adopt in my own games, but it's not strictly an answer to the question, even if it is a way of addressing the problem which motivated the asker. \$\endgroup\$ – Carcer Aug 7 '18 at 13:35
  • 2
    \$\begingroup\$ I was trying to address the poster's point saying: "It makes me wonder if I've overlooked any rules that otherwise balances the options. Am I?" While not an overlooked rule this is one which balances the options. I will make this clear in my answer. \$\endgroup\$ – shadydave Aug 7 '18 at 13:40
  • \$\begingroup\$ That's fair enough. \$\endgroup\$ – Carcer Aug 7 '18 at 13:47
  • \$\begingroup\$ I actually have a similar house rule, where you can re roll 1s and 2s but have to keep the new roll. This does a couple of things, it actually becomes statistically better to roll instead of taking the average (I play with two maths PhDs as players, who are obviously going to choose the option with the highest expected value). It removes that horrible "I rolled a 1" feeling, but doesn't completely eliminate the 1 as an option. I like players having a little variability in HP as it makes for more interesting boons and challenges in play. \$\endgroup\$ – MooseBoost Feb 5 '19 at 14:30

A small thing worth mentioning that I haven't seen anywhere below, and would like to add to the record: Rolling is actually mechanically advantageous (or at least a lot better) for characters with exceedingly crappy hit dice and CON bonus.

For instance, a Wizard with 6 CON would roll d6-2 for his hit points on level up, minimum 1. That means the Wiz will get 1 hit point if he rolls a 1, 2, or a 3, and 2 HP if he takes average.

Instead of the roll distribution being "1, 2, 3, 4, 5, 6," it is now "1, 1, 1, 2, 3, 4." Yeah, the Wizard can take average and still gain two hitpoints instead of one, but from my perspective that's kind of whatever. You might as well roll the die and shoot for the 4. You have a lot less to lose from the roll now.

  • 1
    \$\begingroup\$ Actually, rolling is still worse than taking the average for low con values. In your example, you have a 33% chance to get more HP, a 17% chance to break even, and a 50% chance to lose. This is better odds than rolling if you have CON >9, but still statistically worse than taking the average. \$\endgroup\$ – Shem Feb 22 '17 at 18:22
  • 1
    \$\begingroup\$ @Shem Not so: you're correct that there's a high chance of losing, but you stand to gain more than you can lose. The average of {1,1,1,2,3,4} is exactly 2, so it's mathematically equivalent, and Lucas is correct that to a lot of people, that means you might as well roll the die and shoot for 4. \$\endgroup\$ – SirTechSpec Jun 18 '17 at 0:30
  • \$\begingroup\$ As of the 2018 PHB errata, this answer's claim about there being a minimum of 1 HP gained when rolling the Hit Die upon leveling up is now true. See the answers to this Q&A: Can I have net negative maximum HP per level with a negative CON modifier? \$\endgroup\$ – V2Blast Nov 27 '20 at 6:24
  • \$\begingroup\$ That said, the comparison in the last paragraph of this answer makes no sense - either you need to compare the Hit Die roll alone (which ignores the minimum HP increase of 1), or you need to compare the amount of HP gained from each result of the Hit Die roll (in which case that distribution would never be "1, 2, 3, 4, 5, 6" unless you have a Con mod of 0). Right now, the answer's comparing the distribution of unmodified results on the die to the distribution of possible HP increases from that die roll, which doesn't really make sense as a comparison. \$\endgroup\$ – V2Blast Nov 27 '20 at 6:29

Most other answers attempt to justify the system instead of simply answering your actual question.

None of the highly rated answers state the important basics:

No you aren't missing anything.

You're entirely right - there is no sound reason to roll random hit points, when the deterministic choice nets you a higher average. There is no balancing factor - it's simply worse.

Risk needs to come with reward. In 5th Edition D&D, it's the wrong way round.

Also, the hard cold truth, especially at low levels, is that a low roll is worse for your character than a high roll is good. That is, that 1 on a d12 will hurt your Barbarian far more than the 12 your buddy got is helping him.

Meaning that the risk here is greater than the mathematical average suggests. While you mathematically only lose half a point from taking a risk, in actual play experience this risk is significantly greater. (Though once you're off the lowest levels - around level 5 or so - this factor ceases to be important, simply because once you have 30 or 40 hit points one more or less hp doesn't matter much any longer.)

The default D&D system asking you to choose between taking the "average" and rolling:

  • penalizes players that do not think probabilities through or lack the mathematics training needed (since the statements of this answer are not intuitive, far from obvious, and definitely aren't discussed in the rulebook)
  • penalizes gambling players (since their preferred choice - rolling dice - is inferior)

In short, it's not done right. Any system that exposes you to risk (rolling low) should also come with reward (such as a higher average).

You aren't asking for houserule fixes, but one quick-fix is to deduct 1 from all fixed hit point increases, but not when you roll. I would probably still take the average for my first three levels or so, so it is a very minimal fix.

  • \$\begingroup\$ The reward for statistically lower average is the chance of having more total hitpoints, if you are lucky. Almost half of characters will have more HP by rolling, so I object to saying it is simply worse, only statistically worse. Anybody who has played that 26+ HP 2nd level Barbarian knows this from experience. \$\endgroup\$ – WakiNadiVellir Nov 27 '20 at 5:03

Although the mathematics involved do state that using high-average is a winning option, the game really isn't about winning or losing, thus we cannot simply weigh an argument based on mathematical logic. Each player and GM places their own value on various aspects of the game, and no answer is completely wrong as long as it promotes the enjoyment of the hobby among those playing.

My observation is that the choice of rolling for hit points and stats stems from the early versions of D&D when there was not a choice, and you even rolled for hit points at 1st level. During this time-frame it was not unheard of for a wizard to only have 1 hit point maximum at level 1.

Some players prefer the method because of the risk involved. Some prefer this method due to nostalgia. Some prefer the concept of 'Old-School Gaming'. Some even prefer the grittier game play that the uncertainty can provide. There are enough reasons that are popular enough that it was in WOTC's best interests to keep the option in there.


A) The DM requires it. Sometimes, there are DMs that just want the good old fashioned roll to level system.

B) It's more fun. Many people enjoy the risk they feel whenever they roll the dice.

C) It adds more challenge to the game. Having a slightly weaker character makes the sense of danger more real. Plus, optimization takes longer for the player as well.

D) Nostalgia old-school players like rolling, simply because that's how they've always done it, because it's a critical part of the challenge and achievement, and they remember the good old games.

E) Every hero experiences a rough patch, and the dice help show that. If you roll really low for 2+ levels in a row, it almost adds to the how believable the hero's adventures are. After all, they are mortals, not gods.

  • 1
    \$\begingroup\$ "optimization takes longer for the player as well": that's a good point. The fewer variables that are predictable or under full control, the farther away from a fully-optimised game playstyle the group becomes. (Whether that's a pro or con will depend on the group of course.) \$\endgroup\$ – SevenSidedDie Nov 14 '14 at 22:18

Of course all that's already been said about people liking to gamble is true and the opportunity to gain more than average even if there's in fact a greater possibility to roll less than average is really charming to many people.

It's also true that it was the default system, but this does not really explain why the developers thought it was a good procedure. I speculate it has something to do with rolling a great number of characters in the wargamey many-characters-per-player parties of the first gaming groups: some will have low HP, some will have high and in the end it doesn't really matter because every player could get some squishies and some meaties just like they had some OP casters and some not so powerful mundanes.

I feel like there's another reason, and I also got some extra considerations that might be interesting to you.

I used to roll HP for my games but I faced the same problems you faced: some players were unsatisfied because they rolled bad. One player felt like he always rolled bad at caster level checks or at weapon swings but always had good HD rolls (to the point where rolling particularly high during a level up once caused the death of his character, which again shows how bad it is to trust the dice).
When we started playing D&D 4e everyone was a little bit upset by the lack of HD rolls. In their eyes it was like the character were less alive because they were all the same - every barbarian of any given level was in the same 5hp range as his barbarian comrades.

In my last D&D 3e campaign I used the 5e way to calculate HP and the guys rolling a small dice were very happy. Our wizard invested a lot in constitution and, needing no wisdom periapts, had more HP than both the cleric and the druid. The old average (rounding up the total and not every HD) works better to keep the squishies squishy IMO.

  • 1
    \$\begingroup\$ Sorry but you just piqued my interest. How did he die from rolling too high on his HD? Did his maximum HP overflow and get into negative or something like that? \$\endgroup\$ – El Suscriptor Justiciero Dec 2 '14 at 13:43
  • 1
    \$\begingroup\$ @ElSuscriptorJusticiero He was being attacked by a frenzied berserker right after their target dropped dead. Three high rolls, and he was at -10 with the second. Since the enemy was already dead and the encounter was ended we ruled a level up that could have saved him if the second hit just made him fall unconscious. By rolling max, he went to exactly 0 hp with the second hit and was felled by the third. \$\endgroup\$ – Zachiel Dec 2 '14 at 21:01

Because you believe you are lucky. I have had several players who have believed they were lucky at dice, and some who believed they were unlucky. Although I haven't tested them, a few actually did seem anomalous.

Because your DM allows you to spend inspiration points on the Hp rolls. Not RAW, but I know more than one DM who allows this. In which case while rolls below the average are possible, the expectation is above what you would get for average, rounded up.


Character Evolution: If you create a character who rolls extremely low hitpoints you can just kill him and create a new character, generally at a comparable level with few drawbacks. As a consequence, players who take the average will be worse off than players who roll, after several iterations of characters.

Example: Andrew and Bob both create characters. Bob takes the average of 10d10 rounded up for 60 HP. Andrew rolls and gets 50HP, a terrible roll on average. Both characters get horribly mauled by bears and die. Andrew now re-rolls a new character at the same level and gets 65HP, a pretty good roll. Bob takes the safe option, does not roll, and remains at 60HP. Rolling eventually leads to superior characters in the long run.


The roll option is also to create a lack of uniformity, which from a certain standpoint is more representative of reality as it stands. If for example, two fighters can vary in every single way (different backgrounds, skill choices, fighting styles, race, age) why would it stand to reason that they automatically had the same hit points? Two boxers who follow the same training regime would, technically, have different HP, that ability to stay standing is a major factor in why some win and others lose, as much as skill. Some DMs choose to decide on the 'roll for hit points' to represent this randomness/genetic variation in biological systems between individuals.


There is no mechanical reason to roll the dice, rather than taking the average.

It's basically a DM call, and a difference in how different groups play. In the same way that some groups like to roll for their stats, and some like to use point buy, some groups are going to want to roll for hit points and some won't.

There is a very common house rule from the 3.5/PF era that you can take the average hit points per level, gaining an extra point every other level when the .5s add up. For example, a d4 HD Wizard would get 4 HP at 1st level, then 2 at second, then 3 at third, then 2 at fourth, and so on. It's likely that the "average, rounded up" rule came out of a simplification of this house rule.


Depending on the length and power level of a campaign, a player would roll hit points for his character maybe 10-20 times. This is not a statistically significant number of rolls, meaning you have a perceived chance of rolling over the average.

Just because a d10 averages to a six over tine doesn't mean your tenth level fighter can't have 90 HP (before CON). It's a matter of luck.

By the same token, you can just replace a d20 attack roll with 11, but where's the fun in that?

  • \$\begingroup\$ "you can just replace a d20 attack roll with 11" -- is that something from the playtest that didn't make it into the official release? \$\endgroup\$ – mdrichey Dec 4 '19 at 19:11
  • \$\begingroup\$ No, it isn't,and the reason it isn't is because we want to feel the excitement of rolling to hit, rather than relying on a fixed average "you will always hit/never hit this enemy". By analogy, I advanced the claim that rolling for hitpoints gives you the same feeling of excitement, and the chance to roll higher than the average. \$\endgroup\$ – lisardggY Dec 4 '19 at 19:16

Player efficacy. Or rather, perceived efficacy, if you take the stance that the dice are truly random and not influenced by the dice rolling talent of the player. In my opinion, any game is much more enjoyable when the players understand the intended random nature of the conflict resolution. "The rolls were in my favor tonight!" (sees good fortune in the random rolls) versus "Gee, I rolled good tonight!" (takes ownership of favorable random results). It's better to hone decision making and teamwork than superstitious dice rolling practices, at least in my opinion. But, hey, what's fun is fun.


Although the rules do not say so in 5e (I believe), many groups who have played previous versions of D&D will play with the rule that characters with a negative constitution bonus have a minimum hp gain per level of 1, since that was the rule in every other D&D I'm aware of. (Also, it prevents characters from dying when leveling up).

In this case, for characters with negative con bonus and a small hit die, rolling can have a better average.

Otherwise, no, as every other answer has said.

  • 2
    \$\begingroup\$ For a d4, this is only true if your Con mod is -2. For a d6, d8, or d10, this is only true if your Con mod is -3. For a d12, this is only true if your Con mod is -4. \$\endgroup\$ – Miniman Nov 12 '14 at 23:50
  • 3
    \$\begingroup\$ This is now an official rule, as of the 2018 errata, which adds the bolded portion to the rules on HP increase on level up: "Each time you gain a level, you gain 1 additional Hit Die. Roll that Hit Die, add your Constitution modifier to the roll, and add the total (minimum of 1) to your hit point maximum." \$\endgroup\$ – V2Blast Dec 4 '18 at 21:36

When the issue is ability scores rolling you get:
Is it better to take the array and be Joe Average, or to roll for the odds of getting on average better scores?
For hitpoints you get very opposite answers, it's weird.
Can anyone explain that?
For skillpoints per lvl and spells per lvl and every other creation/advancement stat there is not even a question... hmmm, I wonder why?)

  • Why would I ever choose rolling hit points?
  • Because you are bad at math (and logic).

Joking aside the real answer is you should not. It's sad (but understandable catering) that D&D designers are still offering that option at all. It's demonstrably bad game design (for D&D and similar games) to roll dice for character creation and advancement purposes. Do you roll (in D&D and most similar RPGs) for how many skill-points you get per level or how many spells per level your caster gets or whatever? No you don't. You should not roll for hitpoints (archaic traditions notwithstanding). Rolls should be used for actions (to help DM determine outcomes).

  • But why do people like to roll for hitpoints?

Some people say "oh you have to roleplay the weakness", but would you roleplay a wizard that got less spells per level or a rogue with less skills or whatever? In fact it's the opposite, a good roleplayer will not depend on numbers on a piece of paper, but on the character and personality he has developed for his PC. What is better, to say "my fighter has below average hitpoints so he's really cautious in combat" or "my character lost a comrade and/or half a finger in combat so he's really cautious now"? But people being people will not always choose the optimal or logical path, be that gameplay choices or game design.

There are also a lot of forces of tradition at work in this issue. These forces should not be underestimated. D&D has always suffered from poor design, though 3.0 and 3.5 streamlined and organized a lot of things. But since the default had always been to roll hitpoints (and abilities) from 1st edition (but no other character stats, like thief skills) they kept this option in the default rules for Legacy considerations and catering to some crowds. There doesn't have to be a reason or a valid argument. Legacy alone is enough. (though it's not allowed when there's any competitiveness involved like tournaments or (semi)competitive video games)

  • Again, rolls should be used for actions. (in D&D and similar)

And there will be enough rolls to satisfy anyone's gambling lust in game. Almost every damage done is a roll, so are attacks and saves and much more. From a game design point of view (for games like D&D, task oriented and competitive that need fair character creation) it's "wrong" to roll for hitpoints or anything like that and it's sad (but understandable) that D&D designers are still offering that option at all.

And the same applies to rolling for ability scores. Point buy is fairer. You don't see people rolling for skillpoints per level, do you? Damage is already random/rolled, why would you need random/rolled hitpoints? It's an added level of randomness that is not needed and it's "bad" randomness because it's permanent while damage rolls are temporary.

To understand more about bad randomness and bad game design see KRyan's answer to another question: What effects do critical and fumble tables have on D&D 3.5?

  • \$\begingroup\$ Stop arguing in comments. \$\endgroup\$ – mxyzplk Nov 12 '14 at 18:41
  • 1
    \$\begingroup\$ Certainly true that 3.0/3.5 "organized" rules. But that they "streamlined" them? Given the proliferation of new rules and procedures, that statement would require its own defense. \$\endgroup\$ – Kirt Aug 8 '20 at 16:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.