Ask a casual fan how MMA fights end and the answer is a knockout. Ask the database and the answer is a decision, by a comfortable margin.
One study of 8,591 UFC bouts running through late March 2026 puts decisions at 46.8% of all outcomes, against 32.6% for KO or TKO and 19.3% for submissions. Finishes account for 51.9%, close enough to a coin flip that treating them as the default is simply wrong.
One more figure from that dataset before the divisions: the average fight lasted 10 minutes 37 seconds, while the median was 13 minutes 48 seconds. A mean sitting three minutes below the median means the distribution is skewed by early stoppages, so anyone reasoning about fight length from the average is describing a fight that mostly does not happen.
The Weight Gradient Is Real and Steep
Finish rates climb almost monotonically with weight, and the size of the effect is larger than most people assume.
Heavyweight is the only division where finishing is the norm rather than the exception. Depending on dataset and cutoff, heavyweight stoppage rates land around 66%, with decisions down near 32%. Light heavyweight follows at roughly 61% stoppages and a KO/TKO rate in the mid-forties. Lightweight and welterweight sit close to an even split between finishes and the judges’ scorecards. From featherweight down, decisions win out, settling near 45% stoppages, and women’s strawweight is the extreme case at roughly 66% decisions and a KO/TKO rate around 13%.
That gradient is exactly why method-of-victory odds separate one sportsbook from another. A heavyweight bout and a flyweight bout carry fundamentally different base rates for reaching the cards, and any book pricing both off a generic template is leaving value on the table.
Which means where you place the wager matters nearly as much as which side you take. The Florida sports betting market makes the point cleanly: the state runs regulated online wagering through a single licensed operator under an exclusive tribal compact, so a Florida bettor pricing a heavyweight fight-does-not-go-the-distance line is choosing between one in-state board and the wider offshore market, where fifteen sportsbooks compete on props, bonuses and payout speed. On main-line markets the spread between books is a few cents. On round betting and method of victory it is considerably wider, and that spread is the whole edge.
The Submission Rate Refuses to Cooperate
Here is the finding that breaks the tidy narrative.
Flyweight posts a submission rate of roughly 21.7%, which is effectively identical to heavyweight’s. Knockout power clearly declines as fighters get smaller. Submission threat does not.
The peer-reviewed work on this, published in The Sport Journal, found that for male athletes rising weight class correlates with increased KO/TKO probability and decreased likelihood of both decisions and submissions. The practical result is substitution rather than reduction. Lighter divisions trade knockouts for taps, and flyweight still finishes something close to 46% of its bouts. Smaller fighters are not safer, they are finished differently. Submission rates across the UFC hover near 20% overall and peak in the intermediate classes rather than at either extreme, which is the opposite of the intuition that grappling dominates the small divisions and striking dominates the large ones.
Women’s Divisions Do Not Follow the Rule
This is worth flagging because it is routinely modelled incorrectly.
The same academic work found no reliable correlation between weight class and winning method among female athletes. The authors attribute it to a smaller number of divisions and correspondingly thinner samples rather than to any physiological difference.
The magnitude is worth stating plainly. Men’s flyweight, the lightest men’s division, posts a KO/TKO rate around 23.8%, higher than any women’s division in the same dataset. Applying a men’s weight-to-method curve to a women’s card is a modelling error, not a shortcut.
The Title Fight Trap
The sharpest wrinkle in that 8,591-bout study concerns championship bouts, and it inverts what you would expect.
Title fights finished 57.4% of the time overall, comfortably above the 51.9% baseline. Yet only 46.7% finished inside 15 minutes, against 51.1% for non-title bouts across the same window.
Five rounds rather than three changes the denominator. Championship fights produce more finishes in total precisely because they have ten extra minutes in which to produce them, but they are less likely to end within the three-round window a non-title fight would have had. Anyone treating “does not go the distance” as a single transferable concept across both formats is comparing two different questions.
Finish rates by year, incidentally, show no trend at all. Annual figures from 2015 through 2025 bounced between 44.5% and 52.6% without direction, so the recurring claim that the sport is trending toward more decisions or fewer does not survive the data.
What This Is Actually Good For
Three things follow from all of it.
Base rates belong at division level, not sport level. The aggregate 51.9% finish rate describes no actual division particularly well, and using it as a prior for a specific matchup imports an error of fifteen percentage points or more at either extreme.
Submission threat should be modelled independently of size. It does not decay with weight the way knockout power does, and the intermediate divisions carry more of it than either end of the scale.
Format matters as much as matchup. A five-round bout and a three-round bout are different distributions, and the difference is not a scaling factor.
None of this predicts a single fight. Our rankings exist because outcomes are noisy and one result tells you very little. But the distributions are stable, they differ enormously by division, and they are documented well enough that reasoning from a generic mental model of how fights end is a choice rather than a necessity.