Strength, Dexterity, Constitution, Intelligence, Wisdom, Charisma. Starting values between 3-18.
Don’t count on it. I see several things here that can change.
Ability Score Values
Dungeons & Dragons has used the 3-18 range for over 45 years, at least nominally. Up to Dungeons & Dragons 3e in 2000 you might find a way to exceed the 18 cap (AD&D allowed 18/percentile Strength to fighters, and various other mechanisms could let you get ability scores of 19, 20, or even higher), and in D&D 3e you could have scores higher than 18 if your race supported them, and then go even higher as you gained levels (automatically advance), applied magic (bull’s strength spell or gauntlets of ogre power, or wish for permanent improvement). In D&D 4e if you didn’t start with a score of 20 in one of your main ability scores, you weren’t trying.
Later editions kept the 3-18 nominal range as the base, while allowing characters to go outside that. Emphasis has shifted to highlighting the modifiers rather than the base values, because the modifiers are applied in play much more often than the base values.
Some derivations of Dungeons & Dragons, such as Green Ronin’s AGE System, abandon the base values entirely. In Fantasy Age you still roll 3d6 for each ability score, but that roll us used only to look up the actual score on a table and then discarded.
Ability Score Modifiers
Ability score modifiers have changed over the years as well. In some D&Ds the values are “+1 if 14 or higher, -1 if 7 or lower”, in others they follow the classic BECMI progression (3: -3, 4-5: -2, 6-8: -1, 9-12: +0, 13-15: +1, 16-17: +2, 18: +3… incidentally my favorite) or the D&D 3e-originated formula (base/2-5, rounded down).
Ability Score Options
There are many options when it comes to picking ability scores, and reasons why each option might be right or not.
Classic D&D Ability Scores
We can always fall back on the classics.
Or, if you’re even more ‘classic’ (*cough* old):
They’re a workable enough set, even if their definitions have shifted a little over the decades, and imminently familiar. That familiarity is valuable.
They also measure fundamental aspects of the character. Depending on the edition of D&D we’re looking at, there are varying degrees of simulation involved, and knowing these fundamentals helps us model, in layers of increasing detail, how capable the character is.
Applied Ability Scores
Green Ronin’s AGE System has some ability scores reflecting the classics, but adds some application-specific ability scores.
This can map fairly well onto the classic six. The classic Dexterity is split into Accuracy and Dexterity, Wisdom into Perception and Willpower, Charisma is now Communication (and a bit of Willpower, I think).
Green Ronin’s Dragon AGE uses a slightly different set again.
We could probably remove more fundamental scores and go mostly to how they are applied.
‘Fighting’ could be an ability score measuring natural fighting ability, regardless of how it works. There could affect later decisions about fighting styles: in D&D a ‘tank’ character is likely to have high Strength and Constitution (to hit hard, wear heavy armor, and endure the hits that get through) while a harrier is likely to have high Dexterity (to dodge blows, attack accurately with finesse weapons, and learn two-weapon fighting to increase the number of attacks, and thus damage.
‘Magic’ or ‘Spells’ could similarly determine how good the character is at using certain kinds of magic, with other design options indicating how it works. This could replace high Intelligence for wizards, high Wisdom for divine casters, high Charisma for spontaneous casters.
Discarding Ability Scores
The last few paragraphs suggests another option: discarding ability scores altogether. Characters are all character-shaped, with other mechanisms used to differentiate their abilities.
Hero System eventually discarded the mostly-useless Comeliness score in favor of representing with other elements (Distinctive Features disadvantage, Presence-based powers, and so on).
We could do likewise, probably for all the base ability scores. Humans can lift and carry human-scale amounts, and are human smart, and human nimble, and so on. If someone deviates from this, it could be represented by other means. Rather than each person having distinct ability score values, these differences only come up when they are notable.
In the spirit of keeping things simple to start, and minimizing decisions with small effects. I’m going to go with the last one:
Discarding Ability Scores: Humans are human-shaped. There are no base ability scores at this time. Until other features are added, humans are interchangeable. Differences — between humans or between humans and other — will be done using traits or other features specific to the differences.
If I end up restoring ability scores, I’m likely to explore the Applied Ability Score options. I don’t care if you’re strong, I want to know if you’re good at fighting. Exactly how you’re good at fighting will likely be represented using traits or other features, as described above.
I am likely to discard the ‘base’ ability scores, too. I might use a model like the AGE System, where dice are used to find a lookup value that maps to the actual value. The AGE System lookup table is quite close to the BECMI modifiers above, +1 (i.e. 3: -2, 4-5: -1, 6-8: +0, 9-11: +1, 12-14: +2, 15-17: +3, 18: +4). I’d likely use the BECMI modifiers +1, if I were to do this.