There is a version of MMA history that most fans carry around in their heads. It goes something like this: the early days belonged to grapplers, then strikers caught up, and now everybody is well-rounded. Tidy, and mostly true.
But "mostly true" hides something stranger. We pulled every decisive bout in the MMAHIT archive — 498,372 fights with a winner, spanning from the early 1990s to cards that happened this year — and sorted them by how they ended. The submission is not just losing ground. It has been losing ground every single year for more than a decade, and it shows no sign of stopping.
The numbers, by decade
| Era | Submission | KO/TKO | Decision |
|---|---|---|---|
| 1990s | 47.9% | 29.0% | 14.1% |
| 2000s | 42.5% | 36.6% | 18.0% |
| 2010s | 36.0% | 36.4% | 25.3% |
| 2020s | 29.2% | 36.6% | 32.8% |
Read that first column again. In the sport's opening decade, close to half of every fight that produced a winner ended with somebody tapping. Today it is fewer than three in ten.
Now look at the third column, because that is where the missing fights went. Decisions have more than doubled, from 14.1 percent to 32.8 percent.
And then there is the middle column, which barely moves. Knockouts sat at 36.6 percent in the 2000s and 36.6 percent in the 2020s. Twenty years, and the number is identical.
That is the finding that reframes everything. This was never a story about strikers taking over.
It is not a blip
Decade averages can hide a lot, so here is the same measurement year by year:
- 2013 — sub 38.8% · KO 35.2% · dec 24.5%
- 2015 — sub 36.4% · KO 37.3% · dec 24.5%
- 2017 — sub 33.0% · KO 37.5% · dec 26.1%
- 2019 — sub 30.6% · KO 37.9% · dec 28.0%
- 2021 — sub 29.2% · KO 38.7% · dec 30.5%
- 2023 — sub 30.0% · KO 36.1% · dec 32.4%
- 2025 — sub 28.9% · KO 35.3% · dec 34.4%
Twelve years, and the submission rate falls in ten of them. The two exceptions are tiny. Meanwhile the decision rate climbs in eleven of twelve. You do not get a line that straight from noise. Something structural is happening.
The obvious explanation is wrong
The lazy read is that grappling has declined — that fighters stopped learning jiu-jitsu and started swinging.
The knockout number kills that idea on its own. If strikers were genuinely taking over, KO rate would rise as submission rate fell. It didn't. It sat still for two decades while submissions collapsed and decisions surged.
So the fights are not ending in a different way. They are increasingly not ending at all.
What actually changed: defense became universal
Here is the version the data supports. Submission offense did not get worse. Submission defense got dramatically better, and it got better faster than the offense could adapt.
Think about what a 1995 fighter knew. Many of them had one discipline and a training camp that consisted of their own gym. A wrestler who had never felt a properly applied armbar met a Brazilian who had been drilling it since childhood, and the fight was over in ninety seconds.
That gap does not exist anymore. A journeyman on a regional card in Poland or Ohio today has spent years learning how to defend a rear-naked choke, how to keep a hand in, how to survive a bad position long enough to hear the horn. The floor rose. And when the floor rises, the easy finishes go first — because the easy finishes were never about brilliance, they were about ignorance.
The rounds tell the same story
Break the finishes down by round and the picture sharpens:
- Round 1 — 237,716 finishes
- Round 2 — 84,755
- Round 3 — 26,955
- Round 4 — 1,156
- Round 5 — 432
Two-thirds of all finishes happen in the opening round. Survive it, and your chances of hearing the judges' scorecards go up enormously. That is exactly what a sport with rising defensive competence looks like: fewer fighters get caught cold, and the ones who don't get caught tend to go the distance.
Even the submissions that land have changed
We separated submission finishes by era, and the internal shift is just as revealing.
In the 1990s, the most common submission in our archive was the armbar (592), narrowly ahead of the rear-naked choke (563). In the 2020s the order is not close:
- Rear-naked choke — 19,487
- Armbar — 6,450
- Guillotine — 6,142
- Triangle — 3,261
- Arm-triangle — 2,622
The rear-naked choke now finishes three times as many fights as the armbar. Thirty years ago they were neck and neck.
Why? Because an armbar is usually something you catch — a mistake somebody makes. A rear-naked choke is usually something you build, through takedowns, positional control, and back-taking. As defense improved, the submissions that depended on an opponent's error dried up first. The one that depends on grinding, systematic control survived.
That is the same story again, from a different angle. The sport did not lose its grappling. It lost its shortcuts.
What this means for how you watch a fight
A few practical takeaways from half a million bouts:
The first round is the finish round. If you tune in late, you have probably missed the most dangerous part of the night.
A decision is now the single most likely outcome in modern MMA, at 34.4 percent of 2025 bouts — ahead of knockouts, and well ahead of submissions. Whether you find that boring is a matter of taste. It is definitely a matter of fact.
A fighter with a high submission rate today is doing something genuinely rare. In 1996 a submission specialist was riding a market inefficiency. In 2026 there is no inefficiency left to ride, which makes the fighters still finishing by submission far more impressive than their counterparts from the golden era.
The submission is not dead. But it stopped being the default a long time ago, and the archive can tell you almost exactly when.
Method: figures drawn from the MMAHIT fight archive — 498,372 bouts with a decisive result, plus 13,011 draws and 5,631 no contests excluded from percentage calculations. Bouts are classified by their recorded finishing method. Decade figures cover events dated 1990 through 2026.
