52.6% → 42.3%
win rate by time since the last fight · 14,605 fighter-fights
The naive read: the longer a fighter has been away, the less often he wins. Every figure in it is real, and the conclusion drawn from it would still be wrong.
Layoffs · every UFC fight since 1994
The year
out
Josh Barnett spent 4,179 days out of the UFC, came back, and won inside a round. Ring rust says that should not happen. So we set hundreds of fighters against nobody but themselves — the same man fresh, the same man a year later. Both things turn out to be true.
The question
Is ring rust a real thing — and does it really affect the result?
The received wisdom
A fighter that has a long layoff between fights will be rusty when it comes to competition, regardless of how much they have been in the gym. Do not expect them to shine on their first fight back.
TLDR; Quick Facts
60.3% — 38.2%
share who won their previous fight · short — long layoff
Because a long layoff is not handed out at random. The men who get one are older, and they were already losing before they left. Both predict defeat on their own.
+9.0 pts
792 fighters, each against himself · [+5.9, +12.0]
So compare a man with himself. His own record after a short turnaround, against his own record after a long one: 55.4% falls to 46.4%. The effect survives.
+5.3 — +9.5
penalty after a win — after a loss
And it survives holding the previous result fixed. Even the man who won his last fight pays for the year out, which is what the excuse cannot explain.
+7.5 pts
fighters under 28 · [+0.4, +14.6]
Nor is it age wearing the layoff’s clothes. The penalty is at its largest on the youngest fighters in the sport.
+3.6 pts
losses ending in a finish · [−0.2, +7.4]
But here is what does not change: how you lose. The returning fighter is not getting caught cold. That interval contains zero.
In short
Ring rust is real, and it is worth about nine points of win probability — but not for the reason the phrase implies. The man coming back off a year is not being caught cold and starched; he is being beaten, the ordinary way, a little more often than he would have been. Everything under eight months is indistinguishable from fighting fresh. Twelve is where the record turns.
The number everyone quotes
The cliff at twelve months is not evidence
Yes — by about nine points of win rate once a fighter has been away twelve months, and the effect holds up against every attempt made here to explain it away. That is a rarer outcome on this site than it sounds. Most of what the sport repeats about itself dissolves the moment it is measured properly.
Start with the number everybody would quote, which is real and which proves nothing. Win rate against time since the fighter's last appearance:
Under 4 months
52.6% 3,323 fights · ±1.7
4–8 months
52.2% 7,434 fights · ±1.1
8–12 months
50.4% 2,261 fights · ±2.1
12–18 months
43.8% 982 fights · ±3.1
18 months+
42.3% 605 fights · ±3.9
Flat to eight months, then a cliff. It looks like an answer. It is not one, because nobody is assigned a fourteen-month layoff at random.
Who actually gets a year off
Only 38 in a hundred were winning when they left
A long absence is what happens to a man who got hurt, who is ageing out of the schedule, or who lost and slid down the matchmaker's list. Each of those predicts losing the next fight on its own, with no rust required. And both are visibly true of the group:
Short layoff — under 8 months
- Fighter-fights
- 9,660
- Win rate
- 52.5%
- Average age
- 30.0
- Won their last fight
- 60.3%
Long layoff — 12 months or more
- Fighter-fights
- 1,382
- Win rate
- 43.8%
- Average age
- 31.8
- Won their last fight
- 38.2%
The men coming back from a year out are nearly two years older, and only 38.2% of them won last time against 60.3% of the fighters in regular work. That is not a slightly different population. That is a group already losing before it left, being handed the blame for what happened while it was away.
792 men, measured against nobody but themselves
The same man comes back an underdog
There is one way to settle it. Take only fighters who have done both — who have come back from a long absence and fought on a normal turnaround at some point in the same career — and set each man's record against his own. Same fighter, same chin, same division, same skills. The only thing that differs is how long he was away.
After a short gap
55.4% 5,486 fights · ±1.3
After 12 months+
46.4% 1,246 fights · ±2.8
9.0 points, interval 5.9 to 12.0
The same 792 men, measured against themselves. The interval does not come close to
zero.
Which is also where the number stops being useful. Nine points is a tax on a population, not a sentence on a man — roughly forty-six times in a hundred, the fighter coming back off a year still wins. That is the whole shape of this finding, and it is an awkward one: the excuse is true, and it explains nothing about any particular night.
A fighter who wins a little better than a coin flip in regular work comes back from a year away as an underdog against the same standard of opponent. And the age gap inside this comparison is small — those long-layoff fights come at an average of 32.0 against 30.3 for the short ones, a difference of a year and eight months. The age-band split further down tests that properly rather than asserting it away.
The interval quoted is the unpaired one. On paired data that is conservative: it credits the comparison with less precision than it actually has, which is the safe direction to be wrong in.
Two ways it could still be an illusion
Even the man who won his last fight pays for it
First, the losers. If the whole effect were “men who lose get long layoffs”, then holding the previous result fixed should make it vanish. It does not. Among fighters who won their last fight, a long layoff still costs 5.3 points — interval 0.8 to 9.7. Among those who lost, it costs 9.5 points, interval 5.8 to 13.1. Both clear zero. A man can come off a win, take a year out, and return worse than he left.
Second, age. Splitting by age band keeps the years roughly fixed within each row:
| Age | Short gap | 12 months+ | Gap | 95% interval |
|---|---|---|---|---|
| Under 28 | 59.7% (2,698) | 52.2% (203) | 7.5 | 0.4 to 14.6 |
| 28–31 | 54.2% (3,589) | 46.6% (476) | 7.5 | 2.8 to 12.3 |
| 32–35 | 46.3% (2,426) | 42.0% (433) | 4.3 | −0.8 to 9.3 |
| 36+ | 42.0% (918) | 36.0% (264) | 6.1 | −0.5 to 12.7 |
The penalty is there in every band — 7.5, 7.5, 4.3, 6.1 points — and establishes itself in the two youngest, where the samples are largest. In the older two the interval crosses zero. All four estimates overlap comfortably, so the honest reading is that the cost looks much the same at any age. Those rows carry 433 and 264 long-layoff fights between them. The four estimates all overlap comfortably, so the honest reading is that the cost looks much the same at any age and the older bands simply cannot carry the claim on their own. There is no evidence here that rust spares the young or punishes the old.
The penalty for a long layoff · percentage points of win rate
Same fighter, men +9.0
Clears zero
Same fighter, women +10.1
Clears zero
He won his last fight +5.3
Clears zero
He lost his last fight +9.5
Clears zero
Fighters under 28 +7.5
Clears zero
Fighters aged 32–35 +4.3
Cannot be established
← no penaltybigger penalty →
Each bar is the range the true penalty could plausibly sit in. Cut the data six ways and five of them clear the line — including the man who won his last fight, which is the cut the excuse cannot survive. The women’s bar is the widest here because it rests on 119 fighters against 792, not because the effect is different. Only the 32–35 band fails to clear zero. The 36+ band, not drawn here, does not clear it either. One quantity throughout: percentage points of win rate.
What rust is not
You do not get caught. You get beaten.
The mental picture of ring rust is a man with slow reactions walking onto something — timing gone, chin cold, finished early. The record does not support the picture, only the outcome.
Losses ending by knockout or submission · percentage points, long minus short
Did he lose differently? +3.6
No difference
← no differencecaught more often →
Drawn on the same axis as the six bars above, so it can be compared with them directly: 53.9% of short-layoff losses came by knockout or submission, 2,470 of 4,585, against 57.5% of long-layoff losses, 446 of 776. Five bars up there cleared the line. This one sits on it.
A returning fighter’s losses look like anybody else’s losses. He loses more often, but there is no established evidence he loses differently. Whatever a year away takes, it does not appear to be specifically the chin or the reflexes — it shows up as being second best across the fight rather than as being caught cold in it.
The emblem
Kyoji Horiguchi, nine years later
Horiguchi left the UFC in 2016 and returned on 22 November 2025 — 3,290 days, three days past nine years — and submitted Tagir Ulanbekov. He is the second-longest successful return on record, behind only Barnett.
He is also the shape of the problem with anecdote. Everyone can name a Horiguchi, because the returns that work are the ones that get remembered and replayed. The ones that do not work are a quiet decision on a preliminary card, and there are measurably more of them.
Men and women
The women's game says the same thing, harder
The same within-fighter design runs in the women's divisions on a much smaller base: 119 fighters who have done both. After a short turnaround they won 49.9% of 611 fights; after twelve months or more, 39.8% of 191. That is a gap of 10.1 points with an interval of 2.1 to 18.1.
Wider than the men's, because there is a tenth as much of it — but it clears zero, it points the same way, and its centre sits above the men's 9.0. This is one of the few findings on this site that is independently established in both halves of the roster rather than demonstrated in one and assumed in the other.
What this means for your training
Rebuild the timing, not the chin
A year out is worth about nine points of win probability. That is not a story a losing fighter tells afterwards — it is a number you can apply beforehand, and it points somewhere specific.
-
Live rounds, not sparring wars
There is no established evidence that a layoff gets you finished more often — the gap in losses ending by finish is 3.6 points on an interval running from −0.2 to +7.4, which contains zero. What the numbers describe is a man losing rounds rather than getting caught. That points at timing, distance and fight-shape rather than at durability, and those are rebuilt with live rounds against a resisting opponent, not by proving your chin still works.
-
Eight months is free. Twelve is not.
Everything up to eight months is indistinguishable from fighting fresh. Twelve is where the record turns. If you are choosing when to come back, or advising somebody who is, that boundary is the one worth planning around — the cost is not linear in the time away, it steps.
-
If you are picking the fight, price it in
A returning fighter is not the fighter who left, and the discount is large enough to matter against an opponent in regular work. It applies even to the man who won his last fight before he disappeared — that cut still carries a 5.3-point penalty, which is the one the excuse cannot explain away.
-
Ignore it as a post-mortem
Ring rust is real as a population effect and useless as an explanation of one night. Forty-six times in a hundred the returning fighter wins anyway. Knowing the tax exists tells you how to price a card; it tells you almost nothing about why one particular man lost on one particular evening.
Method: men’s divisions unless stated, full history. A layoff is the gap between a fighter’s consecutive UFC appearances, so a fighter competing elsewhere in between is counted as absent — the database records this promotion only. Debuts are excluded, having no previous fight to measure from.