When creating a character, and specifically when generating its stats by dice-rolling, do you still roll the same number of times when you don't use all the ability scores?

As an example, would a vampire roll to generate 6 ability scores, and then just discard one because it doesn't have a constitution stat; Or, would it only roll 5 scores? This is assuming that the player gets to assign the numbers they roll to the stats they choose.

  • 4
    \$\begingroup\$ If you don't roll for constitution at creation, what constitution does an (ex-)vampire have if they are destroyed and you cast True Resurrection on them? \$\endgroup\$
    – Theik
    Commented Feb 3, 2015 at 9:02

4 Answers 4


The PFSRD's Character Creation Outline lists the steps in order:

  1. Get some Dice!
  2. Get a Character Sheet
  3. Determine Ability Scores
  4. Choose a Race


That list puts determining ability scores (including rolling for stats) before choosing a race. Insofar as the outline would be considered RAW, your vampire would generate 6 ability scores, then choose to play a vampire and lose their CON.

Pathfinder does address creating characters without CON scores, here (scroll down just a bit to the "Construct" type).

That said, as a GM, I would have that character roll one fewer stats than usual (and, in a point-buy game, would require that they leave a virtual 10 there, rather than selling off the points for other stats). This is what Pathfinder says to do.

  • \$\begingroup\$ Advanced Race Guide does allow for the creation of undead characters via creating races. See This Page \$\endgroup\$
    – Aaron
    Commented Feb 3, 2015 at 15:55
  • 1
    \$\begingroup\$ ... so it does, and it explicitly states that Undead characters in Pathfinder follow the "fewer dice" rules. I'll update my answer. \$\endgroup\$
    – minnmass
    Commented Feb 3, 2015 at 19:20
  • 1
    \$\begingroup\$ as a note, it ONLY addresses making characters without CON scores-- if you make a character without, say, a STR score (perhaps by being incorporeal) you still get all your rolls (as explained in part one). \$\endgroup\$ Commented Mar 20, 2015 at 17:44

I am going to go out on a leg here and say, yes, you should, if only because not doing so can lead to really weird situations.

Read this line from the undead type entry:

Resurrection and true resurrection can affect undead creatures. These spells turn undead creatures back into the living creatures they were before becoming undead.

As undead can be brought back from the dead as a living being, not knowing what their constitution score was before they were turned into the walking dead means you won't have any idea what their constitution score would be once they are alive once more.

The downside to this, as I see it, is that you may very well be giving a munchkin the opportunity at a 'free' roll. "Oh, one of my stats is a 5, I'll just add that to constitution because that'll be an ability score - anyway while I am undead" is something I could easily see happening.

If min-max behavior is a concern, it might be preferred to simply assume that they had a perfectly average constitution of 10 and stick with rolling the other 5 ability scores only.


As far as I can tell, Paizo never addressed this. I don't believe Wizards ever did either, so we don't even have them to look to for guidance.

Mathematically, there isn't any good answer, unfortunately. Rolling only five means you have a lower chance of getting a particularly high roll, but also a lower chance of getting a particularly low one. Your results won't have the same probability distribution that the others had. But rolling six gives you the same chance to get a high result, while also letting you just drop your lowest score, which is a large advantage.

Then again, rolling for ability scores inherently initializes your game with imbalanced stats; if balance and fairness are the goals, you shouldn't be using it at all. Point-buy, of course, works out neatly by just multiplying the point-buy value by ⅚. It's not actually fair either way with rolling, but rolling isn't supposed to be fair. Rolling five gives the closer results.


Per the 3.5 srd, looks like vampires have Ø con (http://www.d20srd.org/srd/monsters/vampire.htm). So you'd roll 5.

Vampire Spawn's Ø con link to this http://www.d20srd.org/srd/specialAbilities.htm#constitution which defines Ø con. I found that interesting as well, though a little tangential.

I also believe I've seen monsters with equivalent — scores (though cannot quote any/find any quickly) that would infer the same.

Disclaimer: Very minimal Pathfinder experience but I'd be surprised if a core monster like vampire differed much. Pathfinder seems to have cleaned up some of the stranger 3.5 rules, I would be skeptical this is one of them. Someone with a Pathfinder book (maybe you!) could probably easily tell you if it's Ø/— con or not though.

  • 2
    \$\begingroup\$ Vampires don't have a constitution score of zero, they have no constitution score at all -- which sounds kinda similar, but it's very different. The writing on constitution you've provided isn't special handling of a constitution of 0, it's special handling of no constitution score at all. Someone reduced to 0 constitution doesn't start getting that special handling, as they still have a constitution score (but it's at 0). Note that on the Vampire page you link to, vampires are listed as having "Con Ø", not "Con 0". \$\endgroup\$ Commented Feb 2, 2015 at 23:47
  • \$\begingroup\$ @doppelgreener I've adjusted my 0 to Ø (I admit thinking it a unique font) but note it really doesn't change my answer at all. You'd still roll 5. \$\endgroup\$
    – joedragons
    Commented Feb 3, 2015 at 14:25
  • \$\begingroup\$ Strictly speaking you are still not using the symbol for an empty set, and even if you were I find it a bit weird to use set-notation for what a character's stats are. Not quite as weird as giving it a letter value though. However, it's been a while since I took math, so maybe it does make sense. \$\endgroup\$
    – eirikdaude
    Commented Dec 7, 2015 at 10:43
  • 3
    \$\begingroup\$ A missing stat is usually indicated by a long dash (—). You might want to cut-and-paste that one to use instead of null (Ø) or a bolded hyphen-minus (-), if you're trying to avoid confusion. \$\endgroup\$ Commented Dec 7, 2015 at 21:21
  • 1
    \$\begingroup\$ I have to agree with SevenSidedDie: Use a dash instead of a null symbol. The null symbol is only well-known in a few fields where the concept is important; Most everyone outside of those fields recognises the symbol as a zero (that has been drawn with a slash to differentiate it from the letter O), and zero is precisely what you don't want. \$\endgroup\$
    – GMJoe
    Commented Dec 8, 2015 at 0:23

You must log in to answer this question.

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