Data queried September 18, 2026. We will rerun this page at 250 character sheets.
D&D Beyond can tell you what people build. In its 2023 numbers the Fighter led every class and the Human led every species, as they usually do. What a character builder cannot see is who turned up on game night, or who sat down next to them. Threadfall keeps each campaign's party roster along with the sheets players bring to it, so we counted the characters who are being played, and the parties they are played in.
We read 117 player-character sheets imported into running fifth edition campaigns (from D&D Beyond or a PDF), as they stood on September 18, 2026. Party numbers come from campaigns with at least three active characters on sheets, where active means not dead and not retired. Party size counts every active character in a campaign with three or more sessions, sheet or not. Character counts include the 11 who have since died or retired, because they were played too. A sheet is a snapshot from its last import, so a character who has leveled since is a little behind here. These are self-selected tables on one platform, and nothing below is a claim about all of D&D.
Every party in the data has at least one Cleric, Druid, Bard or Paladin. On its own that is less impressive than it sounds. A third of all characters here play one of those four classes, so five characters drawn at random would include one about 86 percent of the time. The number that chance does not explain is the second healer. Random parties of five would double up about half the time; nearly three in four of these parties do, and about two in three have a Cleric specifically. The forum argument that fifth edition does not need a healer may well be right about the math. These tables bought insurance anyway, and then bought more.
That lines up with what the same tables do in a fight. Our character death numbers show about 15 knockdowns for every death that sticks, and somebody is getting those people back on their feet. One caution: a class that can heal is not a player who does. A Bard can go a whole campaign without preparing a healing spell, and the sheets cannot tell us who spent the slot.
A Rogue shows up in about two in five parties. Someone proficient with thieves' tools shows up in nearly four in five. About one character in four (27 of 115) can work a lock, and two in three of those have no Rogue levels at all: they are Artificers, or they picked it up from a Criminal background. Parties cover the job without hiring the specialist. If you are the GM, a locked door is a fair obstacle at most tables. If you are the player agonizing over whether the group needs you to roll a Rogue, it probably does not.
Only about one party in three has a Wizard. Nearly nine in ten have some full arcane caster, so the seat is filled; it is filled by Sorcerers, Warlocks and Bards. Across all characters, 46 percent play a full spellcasting class, 24 percent a half caster like the Paladin, Ranger or Artificer, and 30 percent a class with no spellcasting of its own. Seven characters in ten can cast something. A GM who expects a spellbook, a ritual for every occasion and a prepared answer to every problem is planning for a caster most parties do not have. The arcane seat mostly belongs to characters who know a dozen spells and cast them with Charisma.
D&D Beyond's 2023 top five classes were Fighter, Rogue, Barbarian, Wizard and Paladin. Here is the main class of the 114 characters at these tables:
| Main class | Characters |
|---|---|
| Ranger | 12 |
| Cleric | 11 |
| Fighter | 11 |
| Sorcerer | 10 |
| Druid | 10 |
| Paladin | 9 |
| Rogue | 8 |
| Warlock | 8 |
| Bard | 7 |
| Barbarian | 7 |
| Wizard | 7 |
| Artificer | 6 |
| Monk, and classes from outside the official books | fewer than five each |
Do not read the order too closely. With 114 characters a gap of one or two is noise, and the top five are tied. Read the shape instead: it is flat. No class tops 11 percent, the Ranger that the internet loves to pity is level with the Fighter, and the Rogue and Wizard sit in the bottom half. Our guess is that a builder counts a lot of first characters and test builds, while a campaign that has run for months is full of people who have already played the Fighter. The Monk is the only class in the Player's Handbook with fewer than five characters.
One character in six (17 percent) is multiclassed, and the most common second class is Fighter.
Humans top every builder count. At these tables they are 12 of 116 characters. Count elves, half-elves, drow and eladrin together and the elf family is the largest group at 25, more than one character in five. After that come Aasimar and Goliath (7 each), Genasi (6), then Dwarf, Dragonborn and Gnome (5 each). Everything else appears fewer than five times, and that includes the Tiefling, fourth on D&D Beyond's list, and the Halfling. Only about half of all characters come from the 2014 Player's Handbook species at all. The 116 characters span more than thirty species, so a GM writing a village that has never seen anything but humans and the odd dwarf should expect the party to be the strangest thing that has ever walked in.
Darkvision is on 65 of the 105 sheets that list senses: 62 percent. The complaint that everyone has darkvision is close, but the party-level picture is more useful. A party where every character has it turns up about one time in six, and a party where nobody does almost never. The typical party is split down the middle. So darkness rarely stops a party and rarely costs it nothing. It divides it: the half that can see scouts ahead, and the half that cannot lights a torch that every creature in the cave can see coming.
We took each character's lowest ability score. Strength is the lowest for 38 percent of them, then Intelligence (24 percent) and Charisma (23 percent). Dexterity, Wisdom and Constitution are each the low score for about one character in twenty. A quarter of these adventurers (29 of 115) have a Strength of 8 or lower; the median Strength is 11 and the median Dexterity is 15.
The highest score tells the other half. Charisma is the most common best stat (26 percent), then Dexterity (23 percent), Strength and Wisdom (17 percent each), Intelligence (13 percent) and Constitution (4 percent). Charisma is near the top of both lists, which matches most tables we have sat at: a couple of players do all the talking and a couple would rather be handed a sword. For a GM, the practical read is that a heavy portcullis is a harder problem for the average party than a locked chest, and a puzzle that leans on Intelligence is not far behind. Median Armor Class is 16. In a typical party the best-armored character sits at 19 and the softest at 13.
82 sheets list an alignment. Chaotic characters outnumber lawful ones better than two to one: 46 percent to 20 percent. Half are good (49 percent), 6 percent are evil, and about one party in six has an evil character in it. Chaotic Neutral (19) and Chaotic Good (18) are the two most common picks. Lawful Good, the paladin on the box art, is 8 of 82.
Perception is on 60 of the 96 sheets that list skill proficiencies, and players spend Expertise on it more than on any other skill. Persuasion (41), Survival (39), Insight (38) and Arcana (36) follow. Performance (8), Nature (11) and Medicine (18) sit at the bottom, so the ambush is the hard thing to pull off and the herbalism puzzle is the easy way to stump a table. Four characters in ten speak Elvish (48 of 117), well ahead of Draconic (25), Dwarvish (22) and Goblin (20), and at least 23 of those speakers are not elves. Villains with a secret should stop whispering in Elvish.
The median campaign has five active player characters. Three campaigns in four sit between four and six, about one in six runs seven or more, and the largest active party is nine. Rosters turn over: of 177 player characters these campaigns have ever had, 39 (22 percent) are dead or retired. The median character is level 5. 41 percent are level 1 to 4, 48 percent are level 5 to 10, and 11 percent are level 11 or higher. The highest is 14, so nobody in the data has reached level 17, where ninth-level spells begin.
| Measure | Value |
|---|---|
| Parties with a Cleric, Druid, Bard or Paladin | all of them |
| Parties with two or more of those | nearly three in four |
| Parties with a Rogue | about two in five |
| Parties with thieves' tools proficiency | nearly four in five |
| Parties with a Wizard | about one in three |
| Characters in a spellcasting class | 70 percent (46 full, 24 half) |
| Multiclassed characters | 17 percent (19 of 114) |
| Human characters | 12 of 116 |
| Characters with darkvision | 62 percent (65 of 105) |
| Lowest ability score | Strength 38 percent; Intelligence 24; Charisma 23 |
| Chaotic vs lawful characters | 46 percent vs 20 percent |
| Perception proficiency | 60 of 96 sheets |
| Active party size | median 5; three in four between 4 and 6; max 9 |
| Character level | median 5; max 14 |
Sheets are the ones players or GMs imported into a campaign's party, from D&D Beyond or a PDF, and each is a snapshot from its last import. Main class is the class with the most levels. Not every sheet carries every field, which is why the counts above run from 82 to 117. Party numbers use active characters with sheets in campaigns that have at least three of them; party size uses every active character, sheet or not. We give party results as rough fractions and do not name any class or species with fewer than five characters, so nothing here points at a single table or a single player. Known confounds: these are tables organized enough to record their games and import their sheets, which leans toward experienced groups; characters whose players never imported a sheet are missing from everything but party size; a healer class on a sheet is not proof of healing at the table; and at this size any gap of a few characters is noise. Test and demo campaigns are excluded. If you want a cut we didn't run, ask: hello@thread-fall.com and we'll run it. The data on this page is licensed CC BY 4.0: reuse it freely, credit Threadfall with a link.
Every party we measured brought one anyway. Each party with three or more active character sheets includes at least one Cleric, Druid, Bard or Paladin, and nearly three in four include two or more. Whether fifth edition can be survived without one is a question these tables left to the forums.
Five. The median campaign has five active player characters, and three in four campaigns sit between four and six. About one in six runs seven or more, and the largest active party in the data is nine.
There is no runaway winner. Among 114 characters in running campaigns, Ranger (12), Cleric (11), Fighter (11), Sorcerer (10) and Druid (10) are in a statistical tie, and no class tops 11 percent. That is a flatter picture than character-builder counts, where Fighter leads by a wide margin.
Strength. It is the lowest ability score for 38 percent of the characters we measured, ahead of Intelligence (24 percent) and Charisma (23 percent). One character in four has a Strength of 8 or lower. Almost nobody dumps Constitution.
About six in ten: 65 of the 105 sheets that list senses. The typical party is split, with half the characters seeing in the dark and half needing a light. A party where everyone has darkvision shows up about one time in six.
It needs the lockpicks more than the class. A Rogue appears in about two in five parties, but someone proficient with thieves' tools appears in nearly four in five. Backgrounds and Artificers cover the job.
Threadfall is the tool that produced this data: it records your session live and turns it into a written chapter. What is Threadfall? · More measured data: how long sessions run · how often characters die · how long campaigns last · how many quests parties juggle · what a real party looks like