Brandon Moreno thought he had won clearly. Then the first scorecard was read.
"In this one I finished, and I felt super confident I got the victory, but then a judge said 29-28 Morales," Moreno said after beating Joseph Morales at Noche UFC in Glendale, Ariz., on Sept. 12, as MMA Junkie reported. His reaction, he said, was "What the fuck?"
The other two judges gave Moreno every round. The press row did not see it that way. MMADecisions, a site that archives official scorecards alongside scores from media members, logged nine media cards for the fight. Six had Morales winning.
So Moreno won a split decision that most reporters thought he lost, and still was not sure the result was safe. "I have bad experiences with decisions," he said.
Some fighters and former champions say the problem is the judges. We checked whether any of them actually stand apart.
The case against the judges
Henry Cejudo, the former two-division champion, went further than most after the UFC 332 main event in October, calling for scoring by artificial intelligence on his podcast. "Too many people have been robbed in fights," Cejudo said, as Bloody Elbow reported. "There's a lot of times where a judge may not like somebody."
His co-host, Kamaru Usman, was not convinced. "I gotta disagree with you there, how is AI supposed to judge this?" Usman said.
Lerone Murphy has his own reason to be unhappy. In London in March, two judges scored their fight for Movsar Evloev and the third had it a draw, giving Evloev a majority decision. Not one of the 17 media scorecards logged for that fight had Evloev winning; most scored it a draw.
Khamzat Chimaev lost his middleweight title to Sean Strickland on a split decision at UFC 328 in May, and he has not let it go. "I didn't lose the match that people think," Chimaev told MMA Junkie. "Like, I never seen a guy win a fight just running backwards and doing just jab, jab."
The press row was nearly as divided as the judges. Of the media scorecards logged for that fight, a narrow majority had Chimaev, and almost as many had Strickland.
What 3,000 decisions show
We took more than 3,000 UFC decisions since 2008 that had at least five media scorecards on record, compared each official card with the media scorers, and looked for the judges who stray from the room.
Most of the time, there is no fight to have. In about nine of every 10 decisions, the official winner was also the fighter most media scorers picked.
The remaining one in 10 is where the arguments live, and those decisions tend to be remembered by the judge's name.
The names do not hold up. Among the judges who scored the most of these fights, the share of their cards that went against the media majority ran from about 7% to 13%.
That looks like a real gap. It is not. A spread that wide is about what chance alone produces when 11 people each score a few hundred fights. Adalaide Byrd had the highest share, but not by enough to say she judges differently from the rest.
Blame the fight
What does predict a disputed result is the fight itself.
Split decisions are where the judges and the room part ways most often. In about one split decision in three, most media scorers had the other fighter.
Chimaev's loss is the pattern in miniature: when the room splits almost down the middle, the judges are scoring a fight with no clear winner. That is what a coin-flip fight looks like. Murphy's loss turned on a first round he called "a toss-up."
"It's a weird fight because nothing much happens in the first round," Murphy told FightersAdvice, as Bloody Elbow reported. "If nothing happens in a round, it's 10-10 or 9-9 or whatever. I think that should be brought in."
He has a point about the system, if not the people. The 10-point must system forces a judge to pick a winner in nearly every round, even one where little happens. Two judges and a room of reporters can watch the same five minutes honestly and land on different fighters.
This year
Disputed decisions have been rarer so far in 2026. Through Oct. 10, the media majority had the other fighter in 11 of this year's decisions.
That is below the long-run rate, but one season is a small sample. A run of knockouts has also meant fewer fights reaching the judges at all.
What to watch
The next time a decision feels wrong, check how close the fight was before checking who scored it. A split decision is a warning that the fight could have gone either way, and the cards usually say so.
Moreno already thinks this way. When a fight goes to the cards, he said, he tries "to be very honest with myself," asking himself, "Maybe this fight is kind of weird – maybe I can lose." Against Morales, he did not think so. Most of the press row did.
Watch the rounds where nothing happens. Murphy's 10-10 idea would change those more than any change of judge.
And watch the AI debate. Cejudo wants a machine to score fights; Usman doubts one can. Our numbers point less at the people in the chairs than at a scoring system that makes them pick a side in rounds that have none.
How we did this
- Data: MMADecisions official judges' cards and media scorecards for UFC decisions from May 24, 2008 through Oct. 10, 2026 (file sha256 11f8821d). We kept decisions with three official judges, a winner, named fighters and at least five media scorecards: 3,109 decisions.
- Media majority: the fighter picked by more than half of all media scorecards for a fight, counting media draws. Under that rule the media majority agreed with the official winner in 2,732 decisions (87.9%), had the other fighter in 308 (9.9%), had a draw in 15 and had no majority in 54. In 41 decisions every media scorecard had the other fighter; in 51, not one media scorecard had the official winner.
- Split decisions: 611 split decisions; the media majority had the other fighter in 200 (32.7%) and agreed with the result in 372. In the 2,498 other decisions, it had the other fighter in 108 (4.3%).
- Judges: 11 judges with at least 150 decisions where the media majority picked a winner. Share of each judge's cards against that majority: Adalaide Byrd 26 of 199 (13.1%), Tony Weeks 35 of 271 (12.9%), Ben Cartlidge 31 of 246 (12.6%), Chris Lee 68 of 552 (12.3%), Sal D'Amato 109 of 966 (11.3%), Michael Bell 57 of 524 (10.9%), Derek Cleary 66 of 620 (10.6%), Eric Colon 40 of 387 (10.3%), Junichiro Kamijo 44 of 426 (10.3%), Ron McCarthy 22 of 264 (8.3%), David Lethaby 13 of 182 (7.1%). A chi-square test of whether these shares differ more than chance would allow: p = 0.56. Judges are not assigned at random, so some judges may see more close fights than others, and the shares are not a strict like-for-like comparison. A judge's own drawn cards are left out; counting them as disagreements puts the shares between 9.0% and 13.9% and the test still finds nothing (p = 0.71).
- By period: 2008-2017, 111 of 1,046 decisions (10.6%) went against the media majority; 2018-2025, 186 of 1,886 (9.9%); 2026 through Oct. 10, 11 of 177 (6.2%).
- Named fights: Strickland vs. Chimaev, May 9: judges 48-47, 48-47 Strickland, 47-48; media scorecards 13 Chimaev, 11 Strickland. Moreno vs. Morales, Sept. 12: judges 30-27 and 30-27 Moreno, 29-28 Morales; media scorecards 6-3 Morales. Evloev vs. Murphy, March 21: judges 48-46, 48-46 Evloev and 47-47; media scorecards 12 draws, 5 Murphy, 0 Evloev.
- Limits: media scorers are not the truth either. They watch from different seats and screens. Agreement with them measures how often judges see a fight the way informed observers do, not whether they are right.