One-Rep Max Calculator
A set taken close to failure, with clean technique.
A label only. Every equation here scales with the weight, so the estimate comes back in whatever unit went in.
Estimated one-rep max
112.5 – 117.5 kg
Four published formulas, disagreeing by 5 kg
- Epley (1985)116.7 kg
- Brzycki (1993)112.5 kg
- Lombardi (1989)117.5 kg
- O’Conner (1989)112.5 kg
How this was worked out
- The set, as read100 kg for 5 repetitions
- Epley, 1985100 × (1 + 5 ÷ 30) = 116.7 kg
- Brzycki, 1993100 × 36 ÷ (37 − 5) = 112.5 kg
- Lombardi, 1989100 × 5^0.10 = 117.5 kg
- O’Conner, 1989100 × (1 + 0.025 × 5) = 112.5 kg
- The range, low to high117.5 − 112.5 = 5 kg, which is 4.4% of the lowest
The four equations
- w x (1 + reps / 30)Epley, Boyd Epley Workout (1985)
- w x 36 / (37 - reps)Brzycki, JOPERD 64(1) (1993)
- w x reps^0.10Lombardi, Beginning Weight Training (1989)
- w x (1 + 0.025 x reps)O’Conner, Simmons and O’Shea, Weight Training Today (1989)
An estimate from one set, not a number to load the bar with unwarmed. Note what happens at a single repetition: Epley and O’Conner read 3.3% and 2.5% high, because neither equation returns the weight on the bar when reps is one. That is the published equation behaving as written, and it is left alone here rather than quietly patched. Every one of these was validated on trained groups performing barbell lifts; accuracy varies by exercise, and LeSuer and colleagues (1997) found all of them underestimating the deadlift. Repetitions are whole numbers here rather than rounded silently: a set of five and a half is a set this page will ask you to resolve before it answers.
Why this shows four answers and refuses to pick one
A one-rep maximum is the heaviest single you could lift today, and the only way to know it is to attempt it. Everything else is a curve fitted to somebody else’s data. Four such curves have been in circulation since the 1980s and 1990s, they were fitted to different lifters on different lifts, and they do not agree. Printing one of them alone would turn a spread of opinions into a fact, so all four appear here with the gap between them measured for you.
Five reps agree, fifteen reps do not
Take a clean set of five at 100 kg. Epley returns 116.7, Brzycki 112.5, Lombardi 117.5 and O’Conner 112.5 — a spread of about 4%, which is smaller than the jump between two plates and small enough to ignore.
Now take fifteen reps at the same weight. Epley says 150, Brzycki says 163.6, Lombardi says 131.1 and O’Conner says 137.5. The gap has grown to 32 kg, roughly a quarter of the lightest answer, and no amount of averaging makes that disagreement go away. Every equation here was built from short, heavy sets; past about ten repetitions they are being asked a question they were never fitted to answer, and the honest response is a range.
| Reps | Epley | Brzycki | Lombardi | O’Conner | Spread |
|---|---|---|---|---|---|
| 1 | 103.3 | 100.0 | 100.0 | 102.5 | 3% |
| 3 | 110.0 | 105.9 | 111.6 | 107.5 | 5% |
| 5 | 116.7 | 112.5 | 117.5 | 112.5 | 4% |
| 8 | 126.7 | 124.1 | 123.1 | 120.0 | 6% |
| 10 | 133.3 | 133.3 | 125.9 | 125.0 | 7% |
| 12 | 140.0 | 144.0 | 128.2 | 130.0 | 12% |
| 15 | 150.0 | 163.6 | 131.1 | 137.5 | 25% |
| 20 | 166.7 | 211.8 | 134.9 | 150.0 | 57% |
The last column is the disagreement as a share of the lowest estimate, and it is the reason this page prints four numbers instead of one. Up to about eight repetitions the four are arguing over less than a plate. By twenty they differ by 77 kg on the same set, and Brzycki is on its way to the discontinuity at thirty-seven reps where its denominator reaches zero and the estimate becomes infinite.
Whose equations these are
Boyd Epley published his poundage chart in 1985 while working with Nebraska athletes: multiply by one plus the reps over thirty. Matt Brzycki gave his in a 1993 article for the Journal of Physical Education, Recreation and Dance, dividing by thirty-seven minus the reps — which is also why it collapses at thirty-seven, where the denominator hits zero. Vincent Lombardi raised the rep count to the power of one tenth in a 1989 textbook, and O’Conner, Simmons and O’Shea added two and a half percent per repetition in a book from the same year.
What a high correlation is not
LeSuer and colleagues tested seven such equations in 1997 across the bench press, squat and deadlift. Correlations were above 0.95 everywhere, which sounds conclusive until you read the next line: every equation understated the deadlift by an amount too large to be chance. A tight correlation means the estimates rise and fall with real strength, not that they land on it. Treat the output as a place to start programming from, then correct it against what the bar actually feels like.
Four equations, machines and reps to failure
Which of the four should I actually train from?
Take the lowest of them for anything you intend to attempt under a bar. The cost of a conservative estimate is one easy session; the cost of an optimistic one is a failed rep with a loaded barbell on your back, and the equations are not precise enough to justify the risk.
Why does a single repetition not return the weight I lifted?
Epley reads 3.3% high and O’Conner 2.5% high at one repetition, because neither ratio equals one when reps equal one. Brzycki and Lombardi return the bar weight exactly. That is a real property of the published algebra, and it is shown here rather than quietly patched over.
Does this hold for machines and dumbbells?
Less well than for barbells. The equations were validated on free-weight lifts where fatigue accumulates in a familiar way, and a leg press or a machine row often allows far more repetitions at the same fraction of maximum, which pushes every estimate upward. Use them for comparison over time on one machine rather than as a true maximum.
How close to failure does the set have to be?
Close enough that one more repetition was genuinely doubtful. If you stopped three reps short, every formula reads the set as easier than it was and understates you by roughly that much again. Consistency matters more than the exact stopping point when you are tracking progress across months.