How Do You Roll Characters?

For an aspect of the game laid out so straightforwardly in the rules, character generation seems to have quite a variety of house rules. How you roll for ability scores is one of the most varied parts of rolling a new guy that I’ve encountered on the internets. (Yes, all of them.)

Standardly, the game has players roll 4d6, drop the lowest d6 and arrange the results of six such rolls as desired. Back in the mists of time the game used to use a straight 3d6, which led to a lot of average characters. This is the reason why no matter what rolling method you use, ability scores have always ranged from 3 to 18. Second edition AD&D had something like five different ability score methods, wherein the DM could pick which one he wanted to use. I think the modern system of “4d6, drop lowest” is fairly elegant.

Still, it’s not without its drawbacks. Most notable is that you’re deciding a lot of your character’s ability and limitations based on a single set of rolls. While most rolls only affect you in the short term—failing a poison save will generally only punish your character for a few days—a lucky or unlucky roll in character generation will last your entire adventuring career. You might create two otherwise identical fighters, one of whom is far more powerful than the other simply because he was lucky enough to roll an 18 or two. It’s exciting and random, prone to creating power gaps between two player characters, and rewards players who cheat (which is easily done). (I once did a statistical analysis of your chance of rolling at least one 18; I’ll have to see if I can find it.)

Point buy is a good solution, with the drawback that it’s not quite as interesting or exciting because it’s not random. However, it’s more tactical—since point buy systems are generally weighted to make high ability scores much more expensive, players often have to choose between a single 18 or multiple 16s. Another drawback is that there’s a lot of futzing about with points and points-weighting charts, and that the standard points value tends to create characters that look a little pale in comparison to the 18-wielding heroes that come with rolling ability scores.

More straightforward are standard arrays of ability scores which are arranged as desired. My own array, designed to allow ridiculously powerful characters to compensate players for losing out on the excitement of rolling, is 18/16/16/14/12/10. Standard arrays are picking up popularity on the internet, making me wonder if I’ve helped propogate the meme in some way. Another popular array is 18/16/14/12/10/8, although I ditched the 8 myself because I don’t like heroes running around being less charismatic than commoners. My own array’s double 16s also reward players who play dwarves —what we used to call demihumans—who can use this array to start with a highly coveted two 18s.

Here are some more specific variants that I’ve seen used:

  • 5d6 drop lowest two is a higher-powered variant on the traditional rolling system. You get a higher distribution overall with this method, so characters tend to be powerful.
  • Reroll ones, an idea that can be applied to any rolilng method, helps prevent the problem of rolling low numbers. Any dice which come up a ‘1’ are rerolled. Generally, ones are only rerolled once each.
  • Points shift is a method my former DM used consistently before he moved away and forgot to get his copy of Return to the Temple of Elemental Evil back from me. This rolling variant allows you to move a certain number of points, usually one or two, from any ability score to any other. Naturally, the minimum of 3 and maximum of 18 (before racial bonuses and penalties) still apply.
  • Iron Heroes points buy is a simplistic, but versatile system—I would expect nothing less from Mike Mearls. Simply put, you begin with ability scores of all 10 and have a certain number of points to raise those on a per-point basis. However, raising an ability score above 15—15 to 16, or 16 to 17—costs 2 points, while increasing a 17 to an 18 costs four points. This is the system I currently use; I offer 26 points, but you may prefer either more realistic or heroic ability scores than my players.

So which should you use? Dice are more exciting, plus it really means something when you roll high ability scores. Points buy is fairer, but can be more time-consuming. Standard arrays are quick and easy, but tend to favour classes that only need one good ability score rather than multiple average scores, and players might crave the excitement of rolling characters randomly. Which on you choose really depends on what kind of game your players will enjoy the most—try to pick the right one for your game.

Comments (32)

ScottM (October 11th, 2006)

The method we use at home is 4d6, reroll ones. In the RPGA games I play every once in a while, point buy (25 or 28 points per the DMG chart) is the rule.

Both of your arrays are far more generous than even the DMG’s “elite array” of 15/14/13/12/10/8.

Jonathan Drain (October 12th, 2006)

ScottM: That’s quite true, I’m over-generous. The initial reasoning was that the players would be quite displeased with the “elite array”, being used to having a chance of rolling eighteens and sixteens.

I’m working on some perl scripts as we speak to calculate statistics on die rolling.

Benjamin (October 17th, 2006)

The system I like to use for starting off a long running campaign is as follows:

-All players gather around to roll ability scores together.
-Split between everyone assembled (including the DM if the numbers don’t work out), 24d6 are rolled, and their 24 results are recorded on a white board.
-The six lowest d6 results are crossed off.
-Each player than uses the same list of the remaining 18 numbers and assigns three to each ability score, using each number only once.
-These 18 numbers are recorded to be used by any characters who might join the game in the future.

I like this method because it is cooperative, balances the characters, maintains the excitement of rolling, and still allows for the customization of point buy.

Benjamin (October 17th, 2006)

P.S. One disadvantage is that the luck of the dice decide whether the party will be high- or low-powered to a certain extent. Really the players and DM should just decide on this beforehand. As is, the system should create PCs a little tougher than usual, however. At everyone’s discretion, how many dice are rolled and discarded can easily be modified to stack the odds in favour of a tougher or weaker party.

ben.c (November 17th, 2007)

I wonder if this will ever be read, considering this post is over a year old at this point.

While dabbling with True20, I came across a house-ruled character stat generation method that I really like. In True20, starting stats are in the range of -5 to 5 (think of it as just the ability modifier, but with a slightly larger range). Now, to generate PCs, you decide on a total bonus, like, +6, to be distributed amongst the stats. Now start each stat out at -2 (for example) and roll Roll 2d6 for each stat, plus a number of d6’s equal to the sum of the ability scores. That means, if you’re going for a sum total of +6, you roll 18d6. Now, for each 1 that comes up, add one to strength, for each 2 add one to dexterity, for each 3, add one to constitution, etc. (or whatever your preferred stat ordering is). If any stat turns out above +5, reroll the excess dice. This guarantees that each player will have the same total bonus, but still allows some randomness in how the stats turn out.

To adapt this to D&D, I wouldn’t bother with 1d6 for every ability score point, but rather would do it by ability score modifiers. Then, I’d roll 3d6 at the end and add 1 point to each ability corresponding to what turns up on the dice.

The down side? Well, 3d6 ends up with a nice distribution centered around 10 and 11, but this system is much more prone to highs and lows. This is a trade-off I’m willing to take, though.

Randy (November 18th, 2007)

InQuest (RIP) had a method I think is pretty balanced: Everybody gets 75 + 1d6 points to distribute among their abilities, with an 18 costing an extra point.

Ben (November 26th, 2007)

The way i run stat creation is 4d6 drop the lowest, but with a few twists.
I allow players to roll 4 sets of stats and pick the set of 6 they are happiest with, no mixing between them.
Also, this is probably a flaw and broken, but as a long time shadowrun player/gm, i believe in rewarding luck. I a player rolls 6,6,6,6 then instead of an 18, they have a 20 for that result.

Wil (November 29th, 2007)

We sit down as a group and roll 4d6, dropping lowest for each ability, discarding sets until we have a set that includes at least one 17, one 16, and “not terrible” other scores. For the most part, we end up with really good scores, but our games tend to consist of fewer encounters with much higher ELs than our party level (usually 3 to 5 encounters per level, rather than the “typical” ~13).

The DM reserves some power over the stats (like asking players to reroll a couple abilities if they get straight 18s, or letting people redistribute points between scores (at a loss; e.g. +1,-2 or +2,-3) within the 3-18 limits).

S. (December 29th, 2007)

The way I have always done it is let players roll 4d6 and drop the lowest one just like the standard PHB method. However, I let them roll seven times and take the highest six as their ability scores. Sometimes, if I am looking to keep the players closer in power to each other, I will also stipulate that the sum of their scores must equal at least 75 or 80, otherwise they may reroll the whole sh-bang until they get there. This generally results in players having strong abilities but also reduces the tendency for one or two players to be severely lower than the others. On the other hand, it still has the excitement of rolling and doesn’t keep every player at some sort of quantitative equality (like in point buy), which I don’t personally like.

J. (February 1st, 2008)

I recently read about a nice method. All the players roll a set of stats using whichever of the normal methods you like best, so 4d6 drop lowest, maybe 7 times and drop lowest, or just 3d6 or whatever. These sets are collected, so you have for example 4 sets of abilities. Then every player picks his abilities from these. This way there is some randomness, but everyone has access to the highest roll if they want it, or a lower roll if it has more extremes (or less) if that’s what they want.

J. (February 1st, 2008)

Clarification: Every player picks his set of abilities from these sets, no mixing and matching allowed.

s. (February 4th, 2008)

A girl I started gaming with recently introduced me to a new method I rather like. We’ve always done the 4d6, drop the lowest, reroll 1’s, arrange as you see fit method. Rerolls allowed. Using the same model (4d6, no 1’s), you roll 6 sets. Arrange them in a matrix. You can choose any six numbers that occur horizontally, vertically, or diagonally. It’s a playful method that gives you 14 choices, without the obsessive rerolling that I myself am prone to (I never do this for deeply involved campaigns, but for fun little one-shots, I’m fairly well known for staying up into the wee hours of the night, rolling 156 sets of 6 until I get a couple 18’s and total bonuses of 16+).

This is fairly high-powered stat-gen, but if you’re doing the sort of campaign where you’re off to save the world, it makes sense that all involved would be a bit above average. If you want it really high-powered, you can easily allow rerolls of the entire matrix.


ktrey (March 13th, 2008)

My method:

All players roll 3d6 x 8, dropping two scores. All players compare scores and decide which scores to keep. All players use these scores to assign to their attributes.

Avix (May 8th, 2008)

The way that I found to be enjoyable for rolling stats is the commonly used 4d6 drop the lowest, 1s rerolled. I then allow characters to roll 1d4 and add that amount to their lowest stat. This way, unless they are horribly unlucky with their rolls, there will hardly ever be a character that has a sub-average score before racial modifiers. I also provide bonuses for luck. 4 6’s rolled qualifies as a automatic 19, 4 1’s rolled (since you can only do that in one roll due to rerolls) nets an auto 20.

Digo (May 20th, 2008)

My system:

Roll 4d6, drop the lowest and reroll 1s. Repeat until you have 12 scores, then pick your 6 best.

Despite how absurdly powerful this is, my group isn’t usually very lucky with rolls. Only half of them will ever get a 17 or 18 out of it. Only once did anyone get two scores above 16. :)

Toxic Rat (August 4th, 2008)

Here’s how we’ve done it. Players choose their class, then roll 5d6 and drop lowest two for their prime requisite score. Then 4d6 for the rest of the scores. Generates a more powerful PC, but then again, the campaign is dangerous enough to warrant it. :)

Kevin (August 8th, 2008)

The method we current use is this….

Roll 4d6, reroll the 1s, drop the lowest

We do this 7 times, and drop the lowest of the 7.

We then roll d4, and add those points however we wish.

I wrote a little program in Visual Basic to do this…. works wonders!! And saves so much time.

Tiorn (December 3rd, 2008)

Why even roll for ability scores?

I’m working on something different that I think will appeal to players and give them a little incentive as well. However, my old circle of gaming friends seem to have went their own ways, so I have doubts about ever giving this method a workout.

Basically, I’m writing up four ability score ‘packages’ for the players to choose from. My goal is to balance these packages in order to make my players really think about what they want to select. I know I’ll probably have to alter these packages to make the decision tougher.

The first package allows for all 18’s on the abilities (pre-race modifiers). The drawback is that the character will have a significant liability or two to balance the package and make the players think twice about it. Such a liability would be something like a (way more powerful) arch nemesis or the character has (and may be continuing to) live(d) a life of slavery.

The second package is for a fairly strong character, with an array for ability scores something like: 18, 17, 16, 15, 14, and 13. Right now, I don’t have a negative aspect of the package assigned to it, but I might add something later. Something like a condition that the character has to struggle with in day to day life, such as alcoholism.

The third package would have an array such as: 15, 14, 13, 12, 10, and 8. This is where things get different… instead of a negative aspect, I’m looking at a positive bonus for the package. Something such as: starting play with a masterwork, mighty, or class/skill appropriate item.

The fourth package would have an array of: 13, 12, 11, 10, 9, and 8. I would continue the positive trend and allow the same type of masterwork, mighty, or class/skill appropriate item, plus, I would throw in a bonus feat.

You can see that my goal is to steer players away from the higher stat characters and toward the lower stat characters. I think it will work, but I can’t know until I put it into action and see if it can even be tweaked for best use. If nothing else, I can just go back to the drawing board or use an older method. I think its worth a try regardless.

Cory (January 28th, 2009)

Here’s one nobody has mentioned yet:

Gambler’s rolls.

Use 4d6 in order, or “reroll 1’s”, whichever you like. I usually do 4d6-L 7 times, drop the lowest overall roll, the rest go in order.

Here’s the fun part:

For every roll the player makes, the GM secretly makes the same roll. Player sees his own dice and then either keeps his dice, or keeps the GM’s dice, whatever that might be!

This makes rolling your stats more fun as it becomes a strategy game - what kind of character can you create under these constraints? Can you create the character you were imagining? You can certainly influence the outcome.

Just based on odds, you might want to take the GM’s roll any time yours is less than 12, to maximize your total score. But maybe you really want a cleric and you roll a 13 wisdom.. are you willing to try for better? Are you willing to accept some other character class if the GM has a 9 for wisdom?

eric (March 1st, 2009)

My main method has been roll 2d6+6, drop the lowest, and arrainge in any order.

For in person games I use this system:

Averages 13 instead of 10.5, so you get some decent hero characters, but you can still have an ability as low as 7 if unlucky. Sometimes I let them roll seven times on 2d6+6 and drop the lowest, the arainge them however so there is less chance of a below 10 score.

For online games I use the Pathfinder RPG heroic point buy.

I have also used a system where you would rank your abilities in importance, then roll a bunch of dice for each. Top stat would get 7d6, next 6d6, then 5d6, 4d6, 4d6 and 3d6, dropping the lowest dice and ending up with 3-18. Good chance of an 18 in his first or second choices. Lots of variance in results. I had a guy roll 7d6 and end up with a 5. Players hate hopeless characters which stopped me from using it.

Even back in the day when I learned AD&D we never used the 3d6 per stat, no reordering, no dropping system. What you get is what you get and you keep it unless you do not qualify for any class. Ugh!

GiacomoArt (May 21st, 2009)

If only the average game designer would pay half as much attention to probabilities and how they play into the human factor as you do for your blog…

Personally, I find that “building” a player character is always superior to “rolling” a player character. The trouble with flirting with those potential 18’s isn’t merely that you could get saddled with a low number instead. Far worse, you could roll up a truly enviable array of stats while everyone else rolls lousy, then the DM has to choose between beefing up the other PCs (devaluing your great rolls in the process) or letting your exceptional character become an object of jealousy and resentment. As you say, these rolls aren’t just a brief walk on the wild side; they become the defining traits of your alter ego’s entire life, and the other players have to live with them as much as you do.

ArkZ (June 2nd, 2009)

roll 4d6 reroll ones once remove lowest die, 7 times remove lowest total

so if your roll 1,3,2,4

you get to re roll that one once

but you have to keep the second roll

so that one could be a 6 on the second roll or a 1 again

as for the seventh you can chose to replace on roll with that one

Mongoose87 (July 9th, 2009)

My DM likes his characters pretty beefy - we play 5d6, drop the two lowest, done seven times, drop your lowest total. It’s been a while since we’ve had a score under 10.

Obahai (September 5th, 2009)

I prefer the 5d6 drop the lowest score, rolled 6 times. Its definitely meant for high powered play, but the players don’t seem to mind. Its not unusual for a player to play a vampire centaur blackguard for no better reason, than it sounds like fun

Andy (September 28th, 2009)

This is a method my group used once when we were still really new to the game. We’d forgotten which dice to roll for abilities, and didn’t feel like looking it up, so we used a d20, reroll the 1, any number of sets. While it was nice to get a couple 20s (without cheating even) for my cleric’s int and wisdom, I later found it’s statistically less likely to get positive modifiers than 4d6.

I really like Benjamin’s 24d6 idea, though. I think I’ll use that next time I DM.

Greypoet (December 9th, 2009)

I like the idea of adding a liability to a roll of 18….something minor…but something just the same…

I get the players to roll 3 columns of 6 totals (ignoring any roll less than a 6)..I then get them to pick the best number off the 3 totals in the top row (six times down to the bottom)…once they have their final 6 scores, they can distribute them to the abilities of their choosing….I find it gives a well balanced set of scores with luck still being a big factor…( one player still ended up with an 8 as a score )…good role players can really act out their weaknesses well

Kim (May 19th, 2010)

My ‘home group’ has been playing for over 20 years now, and we’ve got a lot of home brewed rules and such. In fact, what we play is a 1st edition AD&D homebrew system with skills based (VERY loosely) on the early Pallaidum system, on the world of Harn.

But to roll a character, we go with a VERY heroic stats generator, but we have soem modification to it: we roll 3d6 AND a d8, max out at 18, drop the lowest die. Rolled 12 times, take the 6 rolls you want. (And it’s common we do NOT take all the highest rolls!) You can also only have one 18, and must have 1 stat BELOW 13, which cannot be Comliness. (We use Comliness instead of Charisma, as we believe Charisma should be RPed.)

We also have a secondary set of stats that we use straight 3d6’s for: sight, hearing, touch, taste/smell and voice. (Bards, our class based more like 2nd edition AD&D, get a bonus to their voice from the DM.)

Mirco (September 21st, 2010)

I allow my various players and use this method myself) to distribute 87 points freely (without cost progression) among the attributes. One 18 and one 17 (or 2x 17 instead) and no stats below 7 are allowed.
This allows the players to feel comfortable with the high stat values of their characters and takes away any power gaps.
The gm can improve enemy stats and roll dc to keep the game challenging.

windcaller (September 16th, 2012)

I roll 3d6 12 times and drop the lowest 6. this usually produces satisfactory results.

Nox (January 8th, 2013)

On campaigns I host, I like to follow 4d6 drop lowest, re-roll 1s, and give all players an automatic 18 to replace the lowest result if they did not get one already. This usually gives pretty decent results

Kristen (August 6th, 2013)

I use 3d6 and 1d8, drop lowest, that way 20s are still possible. I also have them roll one more set than the number of stats, and drop the lowest. Not because I like my PCs to be stronger, but because for some reason my party tends to roll rather low.

Rudolph (October 14th, 2013)

I use this for AD&D 2e with Attributes, Skills, Classes, and saves of D&D 3.5. Point buy flat rate 68 points. No stats lower than 5, or higher than 15 at creation. +1 Attribute point at every level. No consecutive attribute gain between levels. In example if you increased STR at second level you could not increase STR again until fourth level. At 4th, 8th, 12th, 16th and 20th level you may chose to increase the Maximum of one attribute, or gain a second Attribute point. If spending a second attribute point you may not spend it in the same place as the other Attribute until two levels later. In example, Rory Le Grand the Quarter-Elf, adds one Attribute point to his WIS at fourth level. He cannot gain another point in WIS until his sixth level.

Comments for this article are closed.