Throughout this article, I’m spotlighting the wonderful old-school minis from Otherworld Miniatures. The shop has now closed down, but you can still find their figures online (eg, here)—and they’re great!
Years ago, I wrote an article about rolling ability scores in 5e which did surprisingly well. Having lately been getting into OSR-style play—specifically, Old-School Essentials—I thought the topic was worth revisiting.
In OSE, of course, random ability score generation isn’t an option; it’s the only way to make a character. And rather than placing your scores to suit your class, you choose your class to suit your scores. That’s what ‘3d6 down the line’ is all about.
As such, this will never be a guide for ‘optimizers’ per se. And anyway, OSR games put much less emphasis on ‘optimization’ than 5e does, mostly because there’s so much less for you to optimize in the first place. Instead, think of ability scores almost like a kind of astrology. You’re reading the tea leaves: interpreting what the rolls tell you and how to make the most of them.
To that end, this won’t be a flowchart of ‘choose this class if you roll X’. Apart from being mathematically ridiculous (I think there are something like 16.8 million arrays we could roll) it would rather go against the spirit of it all. And besides, once you factor in party composition, the challenge of the adventure, even the kind of referee … it just doesn’t matter that much. If you’re an optimizer, it’s time to let go a bit.
3d6 down the line
Old-school games roll ability scores 3d6 down the line. Not ‘4d6 drop lowest,’ not point buy, none of the alternative methods I and others have written about for 5e. You roll 3d6 for each score, in order—and you don’t even get a say about where they go.
That might sound brutal, but you know what? It isn’t—or at least, it’s not as brutal as it looks. But to understand why, you need to understand what ability scores actually do in this game. And compared to 5e, they do a lot less than you’d think.
To start with, consider the modifiers. If you rolled a 9 in 5e, you’d probably be a bit disappointed. But in OSE, 9 is an average roll, giving neither bonus nor penalty—and that range runs from 9 all the way to 12. Any score in that window is genuinely fine. Even at the extreme low end, a 3 only gives you a −3 modifier, not the −4 you’d get in 3rd Edition.
But the bigger point is what ability scores don’t do. And this is where 5e players’ brains need to rewire a bit.
1. Ability scores don’t affect saving throws. With one exception: your Wisdom modifier is applied to saves against magical effects. Otherwise, your saves are determined by class and level. That’s a massive departure from 5e.
2. Ability scores don’t affect spellcasting. No spell attack rolls, no scaling save DCs. A 13 Int magic-user and an 18 Int magic-user cast spells identically.
3. Ability scores don’t affect skill checks, because skills in the 5e sense don’t really exist. There are ability checks, but they don’t come up nearly as much in my experience. (If a 5e player asks me how to make a perception check, they’re often mildly scandalized to learn you don’t: you describe what you’re doing, you listen to the DM, and you react accordingly. Perception is a player skill, not a character skill.)
So what ability scores do is much narrower. Strength and Constitution still do roughly what you’d expect. Dex affects missile attacks and AC (and Initiative, if you’re not using side initiative). Intelligence largely determines your literacy and how many languages you know. Wisdom we’ve covered, and Charisma appears to do relatively little on the character sheet—though, as we’ll see, Charisma has one trick up its sleeve.
In that sense, your ability scores don’t define your character so much as steer you towards one.
There’s also just something about old-school play that wants you to embrace the randomness. Character creation is quick. Characters are more likely to die. The stakes are higher, and the game is more lethal, but it’s better for it because it pushes the challenge back onto the player, rather than onto the character sheet.
There’s an expectation in OSR games that your character’s death might be funny, or tragic, or a bit pathetic. Lean into it!
What’s typical?
The mean-average of a single 3d6 roll is 10.5. So your chances of getting higher (or lower) than that are exactly fifty-fifty. But we can go further:
- Chance of rolling at least one score of 13 or higher: 83.5 percent
- Chance of rolling two or more scores of 13 or higher: 48.8 percent
- Chance that none of your scores are above average: 16.5 percent
- Chance of rolling at least one 18 (or a 3) on any single score: 2.75 percent
Playing around on AnyDice.com, the ‘default’ 3d6 array—the most statistically likely spread you’ll roll—looks something like this:
14, 12, 11, 10, 9, 7
In other words, one good stat, one bad stat, and four average ones. That’s your baseline expectation. 23.4 percent of characters will have this array or better.
How many good stats do you actually get?
As in 5e, some OSE classes are a little more demanding than others. The four basic classes—cleric, fighter, magic-user, thief—technically don’t have any requirements at all. You could have a 3 in every stat and still play one. Other classes are a little more exacting.
I’m going to structure this around how many good rolls you got. This is similar to what I did in my original 5e article.
A quick aside here for OSE newbies: ‘requirements’ and ‘prime requisites’ are not quite the same thing. Every class has a prime requisite, and some have two; it has no mechanical effect other than to determine whether you get a bonus (or penalty) to your XP. Some of the more advanced classes also have requirements—usually a 9 or higher in one or more specified stats. For example, you have to have an Intelligence of at least 9 to play an elf. In the list below, I underline where a stat is a requirement.
- One good stat: fighter (STR); magic-user (INT); cleric, druid (WIS); acrobat, assassin, thief (DEX); half-orc (STR/DEX).
- Two good stats: barbarian (STR/CON, DEX); dwarf (STR, CON); elf (STR, INT); illusionist (INT, DEX); paladin (STR/WIS, CHA); svirfneblin (STR, CON)
- Three good stats: bard (INT, DEX, CHA); drow (STR, INT, WIS); duergar (STR, INT, CON); gnome (INT, DEX, CON); halfling (STR/DEX, CON); knight (STR, DEX, CON); ranger (STR, WIS, CON).
The chance of rolling at least two 9s and a 13 is actually quite high: 81.2 percent. (Why 13? Because that’s the score you need for your prime requisite to actually give you an XP bonus.) But of course, what matters is rolling them where you need them. For example, to get a five percent XP bonus, a bard needs a 13 in CHA—in addition to the 9 in INT and DEX. Only one in seven arrays will give you that (14.23 percent, specifically).
The chance of rolling a 9 and at least two 13s is much lower: 48.3 percent. But again, getting them where you want them is another matter. To get a five percent XP bonus, a drow needs a 9 in INT and a 13 or higher in both STR and WIS—and the chance of rolling that is about one in 20.
The half-elf, as far as I can tell, is unique in requiring two abilities of 9 or higher (CON and CHA) and having two different prime requisites (STR and INT). To get a five percent XP bonus, they would need 13 STR and INT in addition to the 9 in CON and CHA. This crops up roughly once in every 27 stat rolls (3.69 percent of the time). And to get a ten percent XP bonus, one of the prime requisites needs to be a 16 or higher; the chance of rolling this is 1.2 percent, or about one in 83.
Nudging the dice
There are two ways I can think of to tweak your scores after you’ve rolled. Neither makes much of a difference, but they might shift things at the margins.
The first is a rule that often gets overlooked: you can increase your prime requisite by 1 for every 2 points you subtract from Strength, Intelligence, or Wisdom. Two catches. You can’t reduce any of those below 9—so at least one of them needs to start at 11 or higher. And your prime requisite only benefits if you’re exactly one point below the next modifier threshold (ie, 12 going to 13 or 15 going to 16). Otherwise, you’re moving points around for no mechanical gain.
The second is if you’re using the optional rules in the Advanced Player’s Tome for separate race and class. Some races apply modifiers to specific scores: for example, a dwarf increases Constitution at the cost of Charisma. But again, this only really matters if the affected scores are sitting right next to a modifier threshold—and more often than not, they won’t be.
Con, Dex, or both
In Live to Tell the Tale, Keith Amann makes a great point about defensive stats in 5e. Essentially, Constitution and Dexterity do similar jobs: Con determines how much damage you can soak up; Dex (via AC) determines how often you get hit. In an ideal world, you’d want both. But mechanically, it’s a bit redundant: you mostly need to be hard to kill or hard to hit, not both. (The main exception is lightly armoured melee skirmishers who fight with finesse weapons but are frequently on the front lines.)
Anyway, in OSE, it’s moot: your stats fall where they fall. But I still think the framing is useful to think about:
1. Some characters are more likely to be on the front line because of their HD, armour proficiencies, and class features—and they will take more hits as a result. For them, hit points (ie, Con) matters more. Other classes aren’t looking for the front rank and can make do with fewer hit points and focus on not getting hit at all.
2. Speaking of armour: a character who can wear heavy armour, and afford a decent suit of it plus a shield, will get far more AC out of their gear than any Dex score could give them—which frees them up to lean on Con.
So: if you roll a good Con and a middling Dex, you’re more of a damage sponge. You probably want to play a fighter or something similar: wear the heaviest armour you can afford and stand in the front rank. Good Dex, modest Con? Stay back and try not to get hit. A thief, maybe, or magic-user.
Rolled well on both? Congrats.
Rolled badly on both? There are still options; read on.
Don’t ignore Charisma
5e players routinely underestimate the role of Charisma in old-school games, probably because they’re less familiar with the mini-game that is hirelings and retainers—and how important those NPCs can be to the party’s survival.
In 5e, you either need Charisma for spellcasting (eg, sorcerers) or for skills (eg, persuasion). In OSE, neither of these really apply. The main benefit of Charisma is that it determines how many hirelings you can recruit and how loyal they are to you. With good Charisma, then, you can actually afford mediocre rolls elsewhere because you have other people who can make up the difference. You can hire someone to carry your stuff, search for traps, even take the front line. You’re basically outsourcing everything you’re bad at.
Don’t overlook this. It’s one of the most important—and, in my opinion, most fun—parts of the game.
Gold: the seventh stat?
All of which brings us neatly to why starting gold is almost a seventh stat.
Your starting gold can be the difference between a character who walks into the dungeon with only a torch and a spear, and one who walks in wearing full plate, a shield, and a longsword. The latter is a huge advantage in terms of AC, and from the very first session—although of course, by the time you roll your equipment, you’ve already chosen your class.
Putting it all together
So you’ve rolled your six scores, in order, and you’re staring at the array. Now what?
- Take stock. How many scores are 8 or lower? Some classes need 9s, so let those guide you. Elves need INT 9, as do bards, drow, and duergar (who also need CON). Rangers need WIS and CON. Half-elves need CON and CHA. Several classes need DEX (bards, barbarians, illusionists) or CON (dwarves, gnomes, svirfneblin). Halflings and knights need both. Paladins need CHA.
- Read your defensive profile. Look at Con and Dex side by side. Solid Con and middling Dex calls for good armour: fighter, cleric, paladin, most of the demihumans. Solid Dex and poor Con suggests a class that’s best off avoiding the front line: thief, magic-user, illusionist. Both good? The world is your oyster. Both poor? You’re going to be fragile either way, so try to hire some meat shields and play cautiously.
- Identify your best score(s) and check prime requisites. 13 or above earns you at least a five percent XP bonus—and a 16 or higher earns you ten percent. High INT? Consider a magic-user, gnome, illusionist, or elf. WIS? Cleric, druid, paladin. High CHA? Bard. High STR? Lots of options.
- Remember to adjust. If your prime requisite is a 12 or 15 and your STR, INT, or WIS is at least an 11, you can bump up your XP bonus with very few drawbacks. The advanced rules for split race/class can also help sometimes, but remember the level caps for non-humans.
- Don’t forget Charisma. If you rolled a 13 or higher in CHA, that’s useful regardless of your class. Bear it in mind later when you can afford to hire retainers.
- Wait for the gold roll. You won’t know what you can actually equip until you’ve rolled 3d6×10 for starting gold. Some classes depend on it less than others.
- Embrace the randomness. Pick your class and don’t look back. Bad rolls are memorable. A magic-user with a 6 Con who has to be carried out of every dungeon is more memorable than the optimized one you didn’t get to play. The dice told you a story. Your job is to play it.
Ultimately, that’s what ‘reading the dice’ boils down to: not so much building a character as discovering one. Somewhere in those rolls is a character waiting to be played: a thug, a scholar, a coward, a crusader … Finding out who they are is all part of the fun.
To subscribe, click here. If you like what I do, please support me on Patreon or buy my products on DriveThru RPG. You can follow me on Bluesky @scrollforinit and Instagram @scrollforinitiative. And if you want to make my day, you can buy me a coffee here.