Point buy, standard array, or rolled scores

What the three methods share

Strength, Dexterity, Constitution, Intelligence, Wisdom, and Charisma still become modifiers the same way: subtract 10, divide by 2, round down. A 14 and a 15 are both +2 until a bonus pushes the 15 to 16. In the 2024 rules that bonus comes from the background, after the six numbers exist. In the 2014 rules it often comes from the species. Neither bonus is a fourth generation method. You generate the base, then you apply one edition's bonus, not both.

The standard array and the 27-point buy also share a ceiling and a floor that rolling does not. You will not start with a 7, and you will not start with an 18 before bonuses. Rolling can do both before the table has finished snacks. That is the whole disagreement. Everything else is taste and bookkeeping. If you want the arithmetic of the buy, use the point buy calculator and the 2024 point buy guide. This page is about choosing the generator.

The standard array

The standard array is 15, 14, 13, 12, 10, and 8. You assign them to the six abilities in any order. You do not reroll one of them. You do not swap a 12 for a 13 because the class "needs" it. If you want a different shape, you have left the array and you should be on point buy, where a different shape has a price.

Its virtue is speed and sameness. Every character at the table is working with the same raw material. The fighter's 15 and the wizard's 15 are the same 15. Nobody has to apologize for a lucky 18, and nobody has to ask for a mercy reroll after a pile of 10s. Session zero gets shorter. The cost is rigidity. A paladin who wants three good scores will have a 15, a 14, and a 13, and then a gap. A specialist who wanted two 15s cannot have them inside this list, because the list contains one 15.

The array spends 27 points if you price it on the point-buy table: 9 + 7 + 5 + 4 + 2 + 0. It is not "worse point buy." It is one legal point-buy result that the book hands you so you do not have to do the pricing. A table that says "standard array or point buy, player's choice" has allowed the specialist to take two 15s while the player next to them did not think to ask. Decide whether that choice is open.

Point buy

Point buy starts every score at 8 and gives you 27 points. Raising a score from 8 to 13 costs 1 point per step. Raising 13 to 14 costs 2. Raising 14 to 15 costs 2. You cannot buy a 16, and you cannot buy a 7. A 15 costs 9, a 14 costs 7, a 13 costs 5. Three 15s consume the entire budget and leave three 8s. The standard array is the softer alternative. A common specialist buy is 15, 15, 14, 10, 8, 8, which is also exactly 27.

The method rewards a plan. You can put a 13 where a +1 bonus will land, and you can refuse to pay for a 13 that will sit there forever as a +1 you could have had for one point less. It punishes a player who does not want to plan. The guide to the six scores is the planning aid: attack or casting score, then Constitution, then the armor score, then the save you are afraid of, then the dump the party can cover.

House budgets of 30 or 32 points are still point buy if the costs and the 15 cap remain. They are a different game if the table also lets you buy an 18. Say the number out loud. "We play 27, cap 15, then background bonuses" is a rule. "Bring a character" is how one player arrives with an 18, an 18, and a 16 and calls it point buy because points were involved at some stage.

Rolling

The method most tables mean by rolling is 4d6, drop the lowest die, six times, then assign the totals to the six abilities. Each roll is between 3 and 18. The average of 4d6-drop-lowest sits a little above 12, and the spread is wide. One character can open with a 17 and an 18. Another can open with nothing higher than a 13. Both followed the rule. The dice roller can roll a d6 if you want a public roll. Roll four times, drop one, and write every face down. Do not roll in a private tab and report only the highlight.

Variants exist because groups dislike that spread. Some roll 4d6-drop-lowest seven times and drop the lowest total. Some let you move a roll onto a different ability but not reroll. Some require the whole array to be rerolled if the total modifier is under a line they set. Each variant is a house rule. The book method does not include a mercy floor. If you want a floor, point buy and the standard array already have one, and they have it without a negotiation after the dice land.

Rolling in order, Strength first through Charisma last, is a different tradition. It produces characters the player did not choose. It can be a good session-zero game and a bad way to build the healer the party is missing. Do not use it unless everyone agreed to play what they rolled. Discovering that you rolled a wizard's brain onto a character who wanted to be a paladin is the method working, not a problem to fix with a quiet swap.

How far apart characters get

On standard array, before bonuses, the best score at the table is a 15 and the worst is an 8, and every sheet contains both. After a +2, somebody has a 17. Nobody has a 20 at level 1 unless a house rule added more. The party is even enough that a fight tuned to the group will hit everyone in a similar way.

On 27-point buy the ceiling is the same 15 before bonuses, so the same 17 after a +2. The floor is still 8. The difference is distribution. One character can have two 15s and an 8 where another has the smoother array. A specialist is better at the thing they bought and worse at the thing they dumped. The party is even in budget and uneven in shape. That is usually what people want when they say the array feels dull.

On rolling, the ceiling moves. A character with an 18 before bonuses starts at 20 after a +2, which is a +5 modifier at level 1, while a teammate's best score is a 14. The +5 character hits more often, prepares a higher save DC, and also has more hit points if that 18 landed on Constitution. Encounter math that assumes a +3 attack bonus will make one character look heroic and another look ineffective. The ineffective player did not build poorly. The dice built poorly, and the table signed up for that.

What each method does to hit points

Hit points add the Constitution modifier at every level. A standard-array character who puts the 14 on Constitution and a +1 bonus on it reaches 15, modifier +2, unless they put the 15 there and add +1 to make 16, modifier +3. Getting to +3 costs the 15, which is the only high score in the array. The attack or the spell DC pays for those hit points. That trade is visible and fair, because every character faces it.

Point buy can purchase Constitution 15 and another 15, at the price of three weak scores. A fighter can start at +3 Constitution and +3 Strength after bonuses. A wizard can do Intelligence and Constitution. The Tough builds are this case written out to level 5. You cannot stumble into a +4 at level 1. A +4 requires a score of 18 or 19, and point buy will not sell it.

Rolling can put a 16 or an 18 on Constitution before bonuses. A barbarian who rolls an 18 and adds +2 starts at +5 and gains 5 extra hit points every level compared with a +0, on top of an already large die. A wizard who rolls a 10 and an 11 across the top of the array gains 1 per level where the barbarian gains 5, and also has a worse spell DC. If your table wants rolled scores and also wants characters to share a world, talk about this gap before the dice come out. A minimum total, or permission to fall back to the array if the rolls are cruel, is a kindness. It is also an admission that you wanted point buy's floor and rolling's ceiling, which favors the lucky twice.

Choosing for a real table

Choose the standard array when the group is new, when the campaign is short, or when you do not want character creation to take an hour. Choose point buy when the players like building and you want a specialist to be legal without being lucky. Choose rolling when the table enjoys variance and will actually play a low roll instead of lobbying to replace it. The worst option is rolling for the players who asked to roll and point buy for the player who did not like their rolls. That table has no method. It has a negotiation.

Write the choice in the same place you write whether hit points are fixed or rolled. The two decisions are cousins. Fixed hit points plus point buy produces characters a new player can check with a pencil. Rolled hit points plus rolled scores produces characters nobody can rebuild from memory. Either combination can be fun. Mixing "rolled scores, but fixed hit points, but you may reroll a 1 on the hit die, and also you may swap one ability" is how a level 2 sheet becomes unauditable.

For a public one-shot, use the array. You will not have time to compare six rolls per person and you will not have time to repair a character who cannot contribute. For a long campaign with players who have built characters before, point buy is the method that still feels like a choice in month four, because the 8s were purchased on purpose and the 15s were not an accident the table has to design around.

Switching methods later

If the table is unhappy at level 3, you can rebuild with a different method. Do it for everyone, at the same time, and decide what happens to magic items and story facts. Quietly letting one player move from a rolled 12 to a bought 15, while the rolled 18 at the other end of the table stays, increases the spread you were trying to cure.

A new character joining at level 5 should use the method the original characters used, including the same house budget if you changed it. A replacement character on the standard array next to three rolled characters with a +4 and a +5 will feel the gap in the first fight. So will a rolled replacement who arrives with two 18s beside three point-buy specialists. Match the method. Then calculate hit points with the hand method so the new sheet and the old sheets are adding Constitution the same way.

If you are stuck between array and buy and you do not want to roll, play the array. You can always move a future character onto point buy once you have seen which 8 actually hurt. You cannot un-roll a set of dice the table has already agreed to keep.