A single number that captures how a fighter performs in the second half of fights — when output drops, accuracy fades, and decisions get made. Built from per-round strike data across every UFC fight.
Top and bottom 20 among fighters who have been in the cage in the last two years, high confidence only. Updated automatically as new fights are added. Career-aggregate, not weight-class specific.
Read the faders list first. That's where the metric earns its keep. The elite list compresses at the score cap (most "tireless" fighters tie at 140), so it's a directional list of "these are good," not a ranking. We sort ties by deepest sample so the most-confident readings rise to the top.
Search any fighter to see how their output holds up round by round, across their entire UFC career. Each bar is that round's total output — strikes plus grappling — as a share of their round-one baseline. Above the line, they sustain or push the pace; below it, they fade.
This is career-wide, so it won't match the recent-five-fight score in the leaderboards above — it's the whole body of deep-water evidence, not current form. Fighters who finish too early to build a deep-round sample show a finish note instead of a curve.
Every UFC fighter throws strikes at some pace in round one. The question that decides most fights past round two: do they keep throwing at that pace, or does their output collapse?
The cardio score measures consistency of fight output between round one and rounds three and beyond. It looks at three striking dimensions — strikes attempted per minute, strikes landed per minute, total strikes attempted per minute — plus, for grapplers only, a grappling dimension that combines takedowns landed and control time per round. A fighter who sustains or exceeds their R1 pace in their active dimensions scores high. A fighter whose output drops in the deep waters scores low.
The score is reported as a tier word (Tireless, Steady, Tapers, Fades, Collapses) backed by a 0–150 numeric value. The tier is what bettors read; the number is the supporting evidence.
Strikes-attempted, strikes-landed, and total-strikes are counted for any fighter throwing real volume in R1. The grappling dimension is counted only for real grapplers — fighters landing at least half a takedown per round or holding 45+ seconds of control per round in R1. This avoids inflating a striker's score when they shoot a desperation late takedown out of exhaustion.
Across the fighter's last five UFC fights with R3+ data, we sum every minute of R1 and every minute of R3+, separately. From these we compute average per-minute rates in each active dimension.
For each active dimension, the R3+ rate is divided by the R1 rate. Sample-size-aware caps prevent small-sample inflation: 1.5× ceiling for fighters with 60+ minutes of R3+ data, 1.4× for 30–60 min, 1.3× for 15–30 min, 1.15× for under 15 min. Big samples earn the right to differentiate at the top.
Each dimension's ratio is averaged together, weighted by R1 baseline volume. So a fighter who throws 8 strikes/min and 0.4 takedowns/round has striking weighted ~20× the takedown dimension — striker fades can't be hidden by an inflated grappling ratio. The result × 100 is the score, then bucketed into tiers: Tireless ≥115, Steady 90–115, Tapers 70–90, Fades 50–70, Collapses <50.
Not every fighter has the same depth of cardio data. A welterweight with eight five-round wars has a far more reliable score than a heavyweight whose last three fights ended inside two minutes. We surface that with two tiers:
At least 15 minutes of round-three-plus data across recent fights. The score reflects a real pattern, not noise. Treat the tier word as a real signal.
Some R3+ history but a small sample. Score is directional only — flagged as limited on fight cards. Treat the tier word as suggestive, not conclusive.
The fighter rarely makes it past round two. That itself is a signal — they tend to finish or get finished early. We don't show a cardio score in this case.
A fighter cutting from welterweight to lightweight is depleted on fight night. The same fighter moving up to middleweight is rehydrated and fresher than usual. Their cardio looks different at each weight — and the data confirms it.
The cardio score is computed per weight class. When you see a fighter's cardio tier on an upcoming fight card, it's specifically their cardio at that fight's weight class. If they don't have enough deep-water history at that weight, we fall back to their career cardio across all weights and flag the row as limited data.
Fighters who have moved between divisions can have very different scores at each. The weight-class split surfaces a signal that career averages destroy — a fighter's cardio at lightweight tells you almost nothing about their cardio at welterweight.
This page used to claim cardio predicted winners — up to 66.7% when the gap was widest. In August 2026 we tested that properly and it didn't hold up. We've pulled the claim and retired cardio as a pick factor. What follows is what survived.
So cardio no longer moves a verdict. It doesn't appear as an edge factor on fight cards and it isn't part of any pick. If you saw a cardio edge listed here before, that's why it's gone.
What the same testing did find is that cardio predicts how long a fight lasts. When both fighters hold their pace, about 64% of fights go the distance; when one of them gasses, that drops to about 52%. That's a real pattern, and it's a different question from who wins — so we report it as context on a matchup, not as a reason to back someone.
What the metric is genuinely good at: identifying faders. The bottom of the leaderboard names known cardio collapses with real differentiation. The top compresses at the cap because the metric measures consistency, not absolute output — which is also its blind spot: a fighter who is equally inactive early and late scores well without being hard to fight. Read the score next to the pace it's holding.
A fighter who stopped throwing in R3 because they're being controlled or hurt looks identical to one who stopped because they're tired. The score can't distinguish those cases — it just measures output. Read the score alongside the fight context.
A fighter who paces themselves quietly in R1 and then opens up in R3 will look like an elite cardio machine to this metric — even if their R3 pace is just average. Charles Oliveira is a clear example. The metric measures consistency between R1 and R3+, not absolute output. Read the tier alongside the per-dimension breakdown when something looks off.
UFCStats doesn't expose per-round duration on the data we scrape, so we treat every round as a full five minutes. The final round of a finished fight is shorter, which slightly inflates per-minute stats for finishers. Effect is small and applies evenly across fighters.
The score uses a fighter's last five UFC fights with R3+ data. A fighter whose cardio improved or declined recently won't have that change reflected until they've had several R3+ fights at the new level. This is intentional — recent form is more predictive than career averages.