# The Third Scorecard Source: https://fightstats.online/insights/the-third-scorecard Published: 2026-08-07 > 828 times, two judges watched a fight and agreed. The third saw a different fight entirely. This is the UFC's one-in-five problem. --- 828 times, two judges watched a fight and agreed. The third saw a different fight entirely. This is the UFC's one-in-five problem. Two cards agree. The third is the story. - How often do three trained officials watching the same fight disagree about who won it? One decision in five. 828 of 4,095 — which, since 46.6% of all fights reach the cards at all, means roughly one fight in ten in the promotion’s history has been settled by a scorecard at least one official saw differently. How often do three trained officials watching the same fight disagree about who won it? One decision in five. 828 of 4,095 — which, since 46.6% of all fights reach the cards at all, means roughly one fight in ten in the promotion’s history has been settled by a scorecard at least one official saw differently. Every decision in the UFC ends with three pieces of paper. Most of the time they say the same thing. But 828 times — across 4,095 decisions and 8,796 recorded fights — one card has come back different, and the sport has had to shrug and call it a split. That is not a rounding error. One decision in five splits the judges — 20.2 percent of every fight that has ever reached the cards. Nearly one fight in ten, across the entire history of the company, ended with the people paid to watch it unable to agree on what they saw. ### Disagreement, tallied The split decision arrived early and never left. The first one on record came at UFC 9 in 1996 — three years into the company's life. Since then it has scaled with the sport itself: more cards, more rounds, more close fights, more paper for a dissenting judge to sign. Split decisions per year. 2026 is a partial year. 2025's 35 was the lowest full-year count since 2020. ### Detroit, 1996 The first split decision in the data is not a footnote fight. It is Dan Severn against Ken Shamrock — the biggest rivalry the young company had — at UFC 9 in Detroit, May 16, 1996. The rematch went the distance, and when it ended, even the judges couldn't settle it cleanly. Severn took the cards, split. #### SEVERN TAKES SHAMROCK ON SPLIT CARDS — JUDGES DIVIDED FOR FIRST TIME The superfight rematch goes the distance at UFC 9, and the sport discovers a new way to end a fight: in disagreement. The Octagon has produced knockouts, submissions and referee stoppages, but Motor City Madness delivered something new — a decision the judges themselves could not agree on. Dan Severn and Ken Shamrock fought to the final bell, and when the scorecards were read, one dissented. Severn left with the win; the sport left with a precedent. Of the five decisions rendered in the UFC by the end of 1996, this was the only split — the first entry in a column that now runs 828 deep. From the fight database · first split decision on record ### Andre Fili, edge dweller Nobody in UFC history has won more split decisions than Andre Fili. Five times — against Dennis Bermudez in 2018, Charles Jourdain in 2020, Bill Algeo in 2022, Cub Swanson at UFC 303 in 2024, and Christian Rodriguez in 2025 — Fili has stood in the cage while two cards went his way and one went the other. Only two other men have ever reached five: Sean Strickland and Gleison Tibau. Nobody has reached six. Seven years of Fili's career, decided by the one judge who saw it his way. The other side of that edge is just as crowded. Max Griffin, Cortney Casey, Angela Hill, Clay Guida and Paul Felder have each lost four split decisions — careers shaped by the card that went the other way. ### Even with gold on the line The split decision does not respect occasion. Thirty-eight times, a UFC title fight — the biggest, most scrutinised, most consequential kind of fight the company can stage — has ended with the judges divided. Championships have changed hands, or not, on the opinion of one person at cageside. That is the quiet lesson in the 828. The sport has spent three decades getting better at measuring itself — strike counts, control time, knockdowns, all of it tallied to the second. And still, one decision in five ends with trained observers watching the same fifteen or twenty-five minutes and reaching opposite conclusions. The data can count almost everything. It still can't make three people agree on what they saw. Men and women ### The judges disagree at exactly the same rate Two games that share almost none of their statistics, scored by the same officials under the same criteria — and the disagreement rate lands in the same place twice. #### Of decisions are split 3,506 decisions. Men reach the cards in 44.7% of their fights. #### Two tenths of a point apart 589 decisions — from a game that goes to the cards 63.4% of the time, nineteen points more often than the men’s. A difference of −0.2 points, with a range from −3.7 to 3.3. Women’s fights are far likelier to reach a scorecard and far less likely to be settled by a knockout, and none of that moves the rate at which three officials fail to agree. That points at something about judging rather than about fighting: whatever produces a split, it is not a property of the kind of fight being watched. The advice below therefore applies identically to both. What this changes ### Every round you leave ambiguous, you are lending to a stranger Nearly half of all fights reach a scorecard, and one in five of those ends with an official who saw it the other way. Put together, that is the strongest argument this site can make for finishing. #### Anything that removes the judges 4,095 of 8,796 fights have gone to the cards. Preparation built around finishing covers slightly less than half of the outcomes on record; the rest are settled by opinion, and the numbers below say how reliable that opinion is. #### A close round is a coin you handed away 828 of those 4,095 decisions were splits. If a round is close enough that you would need to argue for it, the historical record says there is a real chance somebody at cageside is about to argue the other way — and their opinion is the one that counts. #### That the big fights get watched harder Thirty-eight UFC title fights have ended in a split decision. The most scrutinised bouts the company stages, with the most experienced officials assigned to them, divide the judges too. Occasion does not tighten the scoring. Two honest limits. A split decision records disagreement , not error — it means the fight was close enough to be seen two ways, which is not the same as somebody getting it wrong. And this data counts outcomes, not criteria: it can tell you the risk of leaving a fight to the cards, but nothing here identifies what actually persuades a judge, so it cannot tell you how to win the rounds you do leave. Source: ufcstats.com via Fight Stats · Data as of 7 August 2026 Imagery: Fighter imagery: UFC athlete portrait, mirrored for layout More insights --- ## Data Structured claims: https://fightstats.online/insights/the-third-scorecard/data/findings.json Underlying query results: - https://fightstats.online/insights/the-third-scorecard/data/fili_split_wins.json - https://fightstats.online/insights/the-third-scorecard/data/first_split_decisions.json - https://fightstats.online/insights/the-third-scorecard/data/men-vs-women.json - https://fightstats.online/insights/the-third-scorecard/data/most_split_decision_losses.json - https://fightstats.online/insights/the-third-scorecard/data/most_split_decision_wins.json - https://fightstats.online/insights/the-third-scorecard/data/split_decisions_by_year.json - https://fightstats.online/insights/the-third-scorecard/data/title_fight_splits.json - https://fightstats.online/insights/the-third-scorecard/data/totals.json --- Analysis and prose licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying fight records sourced from ufcstats.com. Method: /data.