Pub quiz rules and round formats: how a quiz night is structured, scored and settled

A pub quiz is not one long list of questions. It is a sequence of short rounds, usually six to eight of them, each with its own theme and its own answer sheet. Ten questions per round is the standard; eight is fine for a themed round that would otherwise run thin. Between rounds the sheets are collected, marked, and the running totals are read out — and that reading out is half the entertainment, so it is worth doing with some ceremony.

Teams are the unit, not individuals. Four to six per team is the range that works: fewer, and a team with one strong quizzer dominates; more, and the people at the ends of the table stop being consulted. Most quizmasters cap teams at six and split larger groups, which is also the only way to keep the room fair when a party of twelve turns up.

A full evening runs about two hours including a break. Six rounds of ten at roughly ten minutes each, a fifteen-minute half-time after round three for the bar and for marking, and a final round with something at stake. Longer than that and the room thins out; shorter and it feels like a warm-up for something that never came.

The default is one point per correct answer, no half points, no partial credit. It sounds blunt, and it is — that is what makes the marking fast and the arguments short. Where a question has two parts ("name the author and the year"), say so up front and make it two points or make it one point for both. Never invent half points at the marking table; a rule that appears mid-quiz is a rule somebody will feel cheated by.

The joker is the one scoring device that is genuinely worth adding. Each team nominates one round before the quiz starts in which their score is doubled. It rewards knowing your own team ("we have a sports person, joker the sports round") and it means a team can come back from a bad start. Two rules keep it honest: the joker must be declared before the round begins, in writing on the sheet, and it cannot be played on the final round — otherwise everybody plays it there and it stops meaning anything.

Some quizzes add a bonus for a perfect round or a penalty for a wrong answer. Both are worse than they sound. A perfect-round bonus punishes the team that missed one question by two points instead of one, which nobody experiences as fair. A wrong-answer penalty makes teams leave answers blank, and a quiz where people stop guessing is a quiz where nothing happens.

General knowledge is the opener: it gets every team on the board and nobody feels out of their depth. After that, alternate a specialist round with a broad one so the lead changes hands. Geography, history, science, sport, music, film and television are the reliable specialist themes; a good quiz has three or four of them and never the same four two weeks running.

The picture round is the one teams remember. Ten images — logos with the name removed, famous faces, landmarks from an unusual angle, album covers with the text blanked — handed out on paper at the start and collected at half-time. It runs in the background while the spoken rounds go on, it gives the quiet teams something to do during the break, and it rewards a different kind of knowledge than the questions do. The music round is its cousin: ten clips of a few seconds each, name the artist, name the song. Both need preparing in advance and both are worth it.

What does not work: rounds with a single answer that either everyone or no one gets, "guess the year" rounds scored as right or wrong (score those by distance instead — see the next section), and any round whose questions the quizmaster has not tried on somebody else first. A round that seemed hard when it was written and turns out to have nine gimmes is a round that decides nothing.

Two teams level after the last round is not rare; it happens roughly one quiz in five. The clean way to settle it is a single estimation question, decided by whoever is nearest: how many steps in the Eiffel Tower, what year the pub was built, how far is it from here to the coast. Nearest wins, no second question, no sudden death. It takes thirty seconds and everybody accepts it, because being nearest is visibly fair in a way that "one more general knowledge question" is not.

Keep two tie-breakers written down before the quiz starts, because the first one may be needed for third place too. Make them questions with an answer nobody in the room could know exactly — the point is the estimate, and a question somebody happens to know is a coin toss for everyone else. If you want estimation as a whole round rather than only a tie-breaker, score it by distance: closest to the truth gets three points, next two, next one. It makes a good final round precisely because the scores can still move.

The first dispute is spelling. Rule: the quizmaster decides, and the standard is "would an unrelated person reading this know what you meant". "Shakespear" is Shakespeare; "Shakespeare" written next to a different play's title is not. Say the rule at the start and apply it the same way to every team.

The second is the alternative correct answer — the question asked for the capital of Australia, a team wrote Sydney, and another team points out that Sydney was never the capital but Melbourne was, briefly. Rule: the quizmaster's answer stands unless it is actually wrong, in which case every team that gave the truly correct answer gets the point, and the quizmaster owns the mistake out loud. Nothing loses a room faster than a quizmaster defending an answer the room can check on their phones.

The third is the phone. Rule: phones on the table, face down, from the moment the round starts until the sheets are in. A team seen looking one up forfeits the round — the whole round, not the question, because the question cannot be identified. Say it once at the start, cheerfully, and then enforce it once, and you will not have to enforce it again. When the quiz runs on the players' own phones, as it does online, this rule inverts: the phone is the answer sheet, and the clock is what stops the search — twenty seconds is enough to answer and not enough to look up.

Almost all of this carries over to a quiz played over the internet, with three differences. The clock replaces the answer sheet: each question runs for a fixed time and closes, so there is no collecting and no marking — the score updates itself. The joker and the picture round have to be reproduced by the host: run a themed round as its own game and double it by agreement, and use a community quiz with image questions for the picture round. And the tie-breaker becomes an estimation question in the game itself, scored by distance automatically.

The host's job changes most. In a pub the quizmaster reads, marks and referees. Online, in a quiz-night mode where the host does not play, the host reads the question off a shared screen, watches the "eight of ten answered" counter, reveals when the room is ready and moves on. Marking and refereeing are gone; what is left is pacing and commentary, which is the enjoyable part. The guide to hosting an online quiz night covers the room size, the rounds and the cameras in detail.