How to split a group into fair teams

The short answer

Fair team splitting has two parts: random order and even sizes. Random order means nobody is chosen first or last because of popularity, and even sizes mean no team plays with an extra person. Shuffle the full list, then deal the names round robin: first name to team 1, second to team 2, and so on, starting again at team 1 when you run out of teams. With 13 people and 3 teams you get 5, 4 and 4, which is as even as the numbers allow.

Captains picking players in turn is the method most of us grew up with, and it is the least fair one. It ranks people in public and gives the first captain an advantage. A shuffled list removes both problems, and if the result looks lopsided you can reshuffle in the open where everyone can see it.

Random or balanced: which do you need?

Pure random teams are the right choice when the goal is mixing: a class that should work with new classmates, a workshop where departments should meet, a pub quiz or a game night. Over many rounds random teams are fair to everyone, even if a single draw happens to put the three strongest players together.

Balanced teams matter when the result of one match or project counts: a five a side game where one strong team makes the evening boring, or graded group work where one group would get all the top students. Then you add a small amount of structure (tiers or a snake draft) and keep randomness inside each step.

• Mixing and getting to know each other: pure random
• Casual sports, one game: random, reshuffle if clearly uneven
• League play or graded projects: tiers plus random
• Pairs for peer review: random pairs, a new draw every session

How to split with the generator on this page

Paste the names into the Names or options box, one per line. Under Split by choose Number of teams when you know how many groups you need, for example 4 tables or 2 sides. Choose People per team when the size matters, for example pairs or groups of 3. Then press Make teams. The teams appear as cards with the number of members on each card, and the button changes to Shuffle again for a new draw.

The shuffle uses the secure random number generator built into your browser, so every arrangement is equally likely. Copy puts the teams on your clipboard as plain text for a chat or a document, and Download CSV gives you a file with a Team and a Member column that opens in Excel or Google Sheets. Your list stays in the browser for next time and is never uploaded.

Team sizes for common group counts

When you split by number of teams, sizes can differ by at most one person. When you split by people per team, a small remainder is spread over the existing groups instead of creating a tiny extra group: if the leftover people are at most half a group, they join other groups; if they are more than half, they form one more group. That way nobody is left alone in a group of one. The table shows what that means in practice.

Balancing skill without losing fairness

The tier method is simple and transparent. Sort the players into tiers as large as the number of teams: with 4 teams, the 4 strongest players form tier A, the next 4 tier B, and so on. Then split each tier at random into 4 teams of one. Every team gets exactly one player from each tier, and who goes where is still decided by chance. In the generator, paste tier A, set Number of teams to 4 and press Make teams, then do the same with tier B and combine the cards with matching numbers.

The snake draft is the classic alternative when two captains or teachers pick: team 1 picks first, team 2 picks twice, then team 1 picks twice, and so on (1, 2, 2, 1, 1, 2). It cancels most of the first pick advantage. Do the picking in private if possible, so nobody hears that they were chosen last.

Skill is not the only thing worth balancing. For school projects, spread out the students who tend to lead; for sports, goalkeepers or tall players; for workshops, one person from each department. Any attribute can become a tier: put all goalkeepers in one list and deal them out first.

Avoiding the same groups every time

Random does not mean different. If 24 people are split into 4 teams of 6, any two specific people land on the same team with a probability of 5 in 23, about 22 percent. Over five sessions, most pairs will meet at least once, and a few will meet three or four times just by chance. That is normal randomness, not a broken shuffle.

If repeats are a problem, draw a new split and look at it before you announce it. For a small class it is fine to swap one or two names by hand after the draw, as long as you do it openly. For rotating pairs over many sessions, a fixed rotation schedule (round robin pairing) guarantees that everyone meets everyone once, which a random draw cannot promise.

Common problems and fixes

• Someone arrives late: add the name and shuffle again before anyone starts, or add the person to the smallest team.
• Two people must not be together: draw, check, and reshuffle if they ended up on the same team. With 4 teams that happens about one time in four.
• Duplicate or empty lines: the counter under the box shows how many entries were found, so check it against the number of people in the room.
• Teams of one: use People per team rather than a large team count for small groups; leftovers then join existing groups.
• People suspect the draw: share your screen or project it and press Shuffle again in front of everyone.