Swiss pairing explained with a worked example

The Swiss system is a tournament format that pairs players with similar running scores each round while eliminating no one, so a large field can be ranked fairly in far fewer rounds than a round-robin would need [1]. Every entrant plays the same fixed number of rounds, and the pairing method keeps steering players toward opponents on their own score, so leaders meet leaders and the final standings reflect who beat whom rather than who survived a bracket.

The name comes from a tournament held in Zurich in the late nineteenth century, and the format spread because it solved a real problem: a round-robin, where everyone plays everyone, is fair but needs one round fewer than the number of players, which is impossible for an open event with a hundred entrants and a weekend to fill [1]. The Swiss system trades exhaustive fairness for scalability. You do not learn how every player compares to every other player, but after enough rounds the top of the table has genuinely been tested against strong opposition, because winning keeps raising the strength of your next opponent.

The current FIDE Swiss rules are the versions effective from 1 February 2026: the revised Basic Rules for Swiss Systems (C.04.1) and the revised FIDE Dutch System (C.04.3), which FIDE confirmed in a March 2026 reminder to organizers and arbiters [7]. Older rule texts are still easy to find online, so check the effective date on any Handbook page before relying on it.

After each round the arbiter sorts every player into score groups: all players on the same score sit in one group, and pairing happens inside each group from the top of the table downward [1]. Under the FIDE Dutch system the players in a bracket are ranked by tournament pairing number (TPN) in ascending order, the bracket is split into an upper half and a lower half, and the top player of the upper half is paired against the top player of the lower half [6]. Other FIDE-approved systems (Dubov, Burstein, Lim) pair differently, so treat this as one system's shape, not the only Swiss algorithm.

The worked example below is a labelled teaching illustration, not a record of a real event. Eight players carry seeds A to H, where A has TPN 1 (the top seed) and H has TPN 8. Colours start from a drawing of lots before round one: the higher-ranked player of each pair receives the drawn initial colour (White here) when their TPN is odd, and the opposite colour when it is even [6]. We paired all three rounds by hand against the FIDE Dutch absolute criteria and then ran a short script that checks every round for rematches, colour differences outside minus two to plus two, and the same colour three times in a row; the sequence below passes all of those checks. It is a legal pairing a Dutch engine could produce, not a claim that it is the only output the engine would give.

Round one splits the field into an upper half (seeds A to D) and a lower half (seeds E to H), then pairs one against five, two against six, three against seven and four against eight: A plays E, B plays F, C plays G, D plays H. Suppose every upper-half seed wins. After round one, A, B, C and D have one point and E, F, G and H have zero. Round two now has two score groups. In the one-point group the Dutch split pairs A against C and B against D; in the zero-point group it pairs E against G and F against H. Say A and C draw, B beats D, E beats G, and F beats H. Scores are now B on two; A and C on one and a half; D, E and F on one; G and H on zero.

Round three shows the two standing adjustments that make a Swiss more than a simple split. B leads on two points alone, so B cannot be paired inside an empty score group and floats down to the one-and-a-half group to meet A, who has not yet played B. That leaves C in the one-and-a-half group with no partner at that score, so C floats down into the one-point group and is paired against D. This downfloat, a player moved to the next bracket because an odd count left them unpaired, is decided by the Dutch ordered criteria that minimise the number and scores of floaters [6], not by a blanket lowest-player rule. The remaining one-point players E and F pair each other, and the zero-point players G and H pair each other. Every round-three pairing avoids a rematch: B had never met A, D had met B but not C, E and F had not met, and G and H had not met.

Colours follow the same criteria. A player's colour difference is their number of whites minus blacks, and it must stay between minus two and plus two, with no player taking the same colour three rounds running [5]. In the example D arrives at round three on two blacks (colour difference minus two), which is an absolute preference for White, so D takes White against C. H arrives on two whites (plus two), an absolute preference for Black, so H takes Black against G. The draw between A and C in round two, and the float of C in round three, are the parts a beginner misses when pairing by hand, which is exactly why the script that checks them earns its place before anyone calls a sequence correct.

So the priorities the pairing follows, in order, are: never repeat a game, keep players paired within their own score group where possible, split each group into halves and pair across them, respect the colour rules covered next, and only then float a player up or down. Doing this by hand for eight players across three rounds is manageable; doing it for two hundred players across nine rounds is not, which is why organizers run endorsed software rather than pencil and paper. Understanding the mechanics is what lets an organizer answer a player who asks, correctly and calmly, why they were paired the way they were.

1. 1Sort all players by score into score groups, highest score group first.
2. 2Rank the players inside each bracket by tournament pairing number (TPN) in ascending order [6].
3. 3Split the bracket into an upper half and a lower half of equal size.
4. 4Pair the top of the upper half against the top of the lower half, and so on down.
5. 5Transpose players to avoid a rematch, then float a player down to the next bracket by the ordered criteria [6].
6. 6Apply the colour rules and adjust the pairing to satisfy them where possible [5].

The Swiss system tries to give every player a near-equal number of games with white and with black across the event, and it avoids assigning anyone the same colour three times in a row [1]. Because white carries a small first-move advantage, a fair format cannot let one player have five whites while another has five blacks, so colour history is tracked for every player and factored into each round's pairing alongside the score-group rules.

In practice each player carries two colour attributes into a pairing: their colour difference (how many more whites than blacks they have had) and their colour history (the sequence of recent colours). When two players are paired, the system gives white to whichever player is more owed it: the one whose difference leans toward black, or the one who had black most recently. The first round sets the pattern from a draw: the Dutch rules determine an initial colour by drawing of lots before round one, and the higher-ranked player of a pair receives that initial colour when their tournament pairing number is odd, else the opposite colour [6]. Every later round works to bring each player's running colour difference back toward zero. Where the score-group pairing and the colour rules conflict, the software applies FIDE's ordered criteria to decide which small compromise does the least damage [2].

A named edge case shows why colour is not always decided by preference alone. Take seed C in the worked example above. After two rounds C has played White twice (colour difference plus two), so C has an absolute preference for Black in round three. C would ordinarily float into the one-point group and the natural partner by rank might be a player C has already met; the rematch rule forbids that pairing outright, so the system transposes to D instead, whom C has not played. D arrives on two blacks, an absolute preference for White. The two absolute preferences happen to agree, so the pairing gives D White and C Black and satisfies both the no-rematch rule and the colour limits at once. When they conflict instead, the FIDE ordered criteria rank the no-rematch rule and the colour-difference limits above a player's mild preference, so the engine accepts the smallest colour compromise that keeps the pairing legal.

When the number of players is odd, one player cannot be paired and receives a bye for that round. Under FIDE rules the pairing-allocated bye normally scores a full point [5]. The Dutch criterion minimises the score of the bye assignee [6], so it usually lands near the bottom of the standings rather than with a contender, but the recipient is decided by the ordered criteria and is not promised to the same lowest-rated player under every configuration. Eligibility has a hard bar: a player who has already received a pairing-allocated bye, or who has already scored a win's points in a round without playing, cannot receive another one [5].

There is a second, different kind of bye that organizers should not confuse with the first. A requested or half-point bye is one a player asks for in advance because they cannot play a particular round (a late arrival, a clash with school, a religious observance). It is usually worth half a point and is announced in the tournament rules, with a deadline for requesting it. The distinction matters for communication: a full-point pairing bye is something the system allocates, while a half-point bye is something a player elects, and stating both policies plainly in the entry information prevents most of the disputes that byes cause.

Treat the number of rounds as a planning heuristic, not a guarantee. A common rule of thumb is that the field size should not exceed two raised to the number of rounds, which suggests at least four rounds for sixteen players, five for thirty-two and six for sixty-four [1]. That formula only sets the floor at which a single undefeated player is arithmetically possible. It does not promise one clear winner: draws between the leaders, or several players finishing on the same score, are common, which is why tiebreaks exist and why most organizers add a round or two beyond the minimum so late-round games between the leaders actually decide the event.

A quick counterexample makes the point. Six rounds for sixty-four players allow an undefeated run of six wins, but if the two front-runners draw when they meet in the last round, they can finish level on five and a half points each, and the round count has not produced a unique winner at all. The standings are then settled by the published tiebreak order, not by adding rounds. Plan the round count to give the event a credible finish, and rely on tiebreaks, not arithmetic, to separate players who tie.

The round count is really a scheduling decision, because rounds multiply by the time control to fill your day. A weekend open with long time controls might manage five or six rounds in two days, while a blitz event can run nine rounds in an afternoon. Decide the round count and the time control together, sanity-check that the field size supports a decisive finish, and publish the round times before entries open so players can plan travel and byes around them. Our companion guide to choosing tournament time controls covers how the clock setting interacts with the round count, and the swiss rounds calculator turns your expected field size into a suggested number of rounds in one step.

Tiebreaks separate players who finish on the same score, and most of them reward the player who faced tougher opposition on the way to that score. FIDE's Tie-Break Regulations define the standard measures and group them by what they draw on, from a player's own results to their opponents' final scores [8]. Because a Swiss does not have everyone play everyone, two players can finish on the same points against very different fields, so a tiebreak such as Buchholz adds up each player's opponents' scores: beat strong opponents and your Buchholz is higher. Organizers must choose and publish the tiebreak order before the event, because applying it after the fact invites disputes.

The practical rule is to pick a short, published stack: a primary tiebreak, one or two backups, and a final method that can never itself tie (a drawing of lots or board-order rule for the very last resort). Buchholz variants and Sonneborn-Berger both lean on opponents' scores, which is why they can move as later-round results come in, so never announce placings until the final round is complete and all scores are settled. For scholastic and rated events the tiebreak choice also feeds how results are reported upward, and once a coach records a result, the record view makes a student's rating history visible for a family to follow.

app.chesscore.io/students/lucas-m

StudentsLucas M.

PASearch⌘K

AR

Lucas M.

Batch B2 · Thursdays · joined Aug 2025

+118 since June

1342

Academy rating · Lichess games synced

Games · 142Reports · 14FeesAttendance

Won vs. 1410 · French Defense

Best game

Missed fork on move 23

Homework set

Full history kept · survives coach changesEvery change logged · audit trail

Pairings are the responsibility of the arbiter, and in any event beyond a handful of players they are generated by FIDE-endorsed pairing software rather than by hand [2]. The arbiter enters results after each round, the program applies the score-group, rematch, colour, and float rules in the correct order, and the arbiter checks and publishes the output. FIDE's Basic Rules require the pairing rules to be transparent enough that the person in charge of the pairing can explain them [5].

Online platforms automate the same logic end to end. On Lichess, for example, the Swiss tournament format pairs, times, and scores an event automatically, which is why clubs and academies can run internal Swiss events without an arbiter present [3]. Chess.com similarly hosts tournaments that handle pairing and standings for the organizer [4]. Automation does not remove the organizer's job, though: someone still sets the round count, the time control, the tiebreak order, and the bye policy, and someone still answers the human questions when a player disputes a colour or a float. Choosing the tool that fits your event is its own decision, and our guide to chess tournament software compares the options for organizers who have outgrown a spreadsheet.

The Swiss system, round-robin, and knockout are three answers to the same question of how to rank a field, and they trade off fairness, size, and drama differently. A round-robin has everyone play everyone and is the fairest ranking, but it only works for small fields because the round count grows with the number of players. A knockout eliminates the loser of each match and is the most dramatic, but it discards most players quickly and can crown a winner who never met the second-strongest entrant. The Swiss sits between them: it suits events with too many entrants for a full round-robin where eliminating competitors before the end is undesirable, and it ranks the field without needing a round for every opponent [1].

This post is about the pairing mechanic itself: score groups, colours, byes, and tiebreaks. The wider job of actually staging an event, from venue and entries to arbiters, prizes, and the day-of schedule, is covered in our main guide on how to organize a chess tournament, and the specific regulations an organizer must enforce live in our guide to tournament rules for organizers. If your event is for schoolchildren, running a scholastic chess tournament has its own considerations around byes, round length, and parent communication that build on the pairing logic here. For rated events in India, how a Swiss result becomes an official rating is covered in our explainer on the FIDE rating pathway in India.