One-Rep Max Calculator
Enter a weight and the reps you got with it. This estimates the most you could lift once by five formulas actually in circulation — Epley, Brzycki, Lombardi, O’Conner and Wathan — shows you how far apart they are, and turns the result into a training-load table you take to the gym. An advanced option lets you pick which formula that table is built from.
The set you actually did
Everything on the bar, bar included. A 20 kg bar with two 20s a side is 100.
Clean reps taken close to failure. Capped at 12 — past that the formulas are measuring endurance, not maximal strength.
100 kg for 5 reps averages out to 115.6 kg across the three classic formulas, which disagree by 5.0 kgon their own. Two more estimates — O’Conner and Wathan — are compared below.
- Epley
- 116.7 kg
- Brzycki
- 112.5 kg
- Lombardi
- 117.5 kg
- O’Conner
- 112.5 kg
- Wathan
- 116.6 kg
- Average of the three
- 115.6 kg
- Spread between formulas
- 5.0 kg
| Percent of 1RM | Load | Reps you should get |
|---|---|---|
| 95% | 110.9 kg | 3 |
| 90% | 105.0 kg | 5 |
| 85% | 99.2 kg | 6 |
| 80% | 93.4 kg | 8 |
| 75% | 87.5 kg | 10 |
| 70% | 81.7 kg | 12 (extrapolated) |
| 65% | 75.9 kg | 14 (extrapolated) |
| 60% | 70.0 kg | 15 (extrapolated) |
| 55% | 64.2 kg | 17 (extrapolated) |
| 50% | 58.4 kg | 19 (extrapolated) |
- All five formulas were fitted in a low-rep range and get worse the further you are from a true single. At 5 reps the three classic formulas already disagree by 5.0 kg.
- Rows marked extrapolated run past the range the load-repetition relationship was measured in. Treat them as a starting point for a warm-up, not a prescription.
- An estimate is not permission. Whether a true one-rep attempt is safe for you depends on your training history, technique under load, and having a spotter — none of which this arithmetic knows anything about.
This is general information, not medical or coaching advice
This page does arithmetic on a weight and a rep count. It does not know your training history, your injury history, your technique under load, whether anyone is watching you lift, or what your heart and joints will tolerate. An estimated one-rep max is a planning number, not a green light.
If you are new to lifting, returning after an injury or a layoff, pregnant, managing a heart or blood-pressure condition, or unsure whether heavy loading is appropriate for you, talk to a doctor or a qualified coach before acting on anything here. Heavy strength work is one of the few places where “consult a professional” is not boilerplate — a maximal attempt is a genuinely different risk from the set you used to produce the estimate.
The five formulas, in full
Each takes the weight you lifted and multiplies it by a factor that depends only on the rep count. That is the whole model — there is no term for the lift, your bodyweight, your training age, or anything else about you.
Epley = w × (1 + r ÷ 30) • Brzycki = w × 36 ÷ (37 − r) • Lombardi = w × r^0.10 • O’Conner = w × (1 + 0.025 × r) • Wathan = 100w ÷ (48.8 + 53.8 × e^(−0.075r))- w
- Weight lifted, bar included — kilograms or pounds, it makes no difference
- r
- Repetitions completed with that weight, 1 through 12
None of the five converts anything, which is why the unit toggle on this page changes the labels and the plausibility bounds and nothing else. Every formula is a pure ratio: 100 kg for five reps and 100 lb for five reps produce the same multiplier, so kilograms in means kilograms out.
The headline figure is the mean of the classic three — Epley, Brzycki and Lombardi — computed from the three rounded numbers shown beside it rather than from their unrounded versions. That costs a hair of precision and buys something more useful: a reader who adds the three and divides by three lands on exactly the number printed above them. O’Conner and Wathan are newer additions to the comparison list, shown at full precision alongside the original three, but they do not enter that average — five formulas voting is not more honest than three, only noisier, and a shared link built on today’s headline has to keep resolving to today’s number.
The training-load table inverts Brzycki specifically, because it remains the cleanest of the five to invert. Set the load to a fraction p of the max and Brzycki rearranges to r = 37 − 36p, which gives exactly one rep at 100% and ten at 75% — the load–repetition relationship printed in strength-training texts, and the same one the National Strength and Conditioning Association’s own training-load chart publishes (the two track within a rep of each other everywhere they overlap). Epley and O’Conner both invert to zero reps at 100%, which is nonsense; Lombardi does not invert to a tidy closed form at all; and Wathan’s exponential, while it does invert, implies over fifty reps at 50% of the max — a faster blow-up than Brzycki’s own. The table itself runs from 95% down to 50%, is built from whichever formula Advanced options has selected (Epley by default), and flags any row whose inversion implies more than ten reps rather than presenting every row as equally solid.
100 kg for 5 reps, step by step
Epley first: 5 ÷ 30 is 0.1667, so the multiplier is 1.1667 and the estimate is 116.7 kg. Brzycki next: 37 − 5 is 32, and 36 ÷ 32 is 1.125, so 112.5 kg. Then Lombardi: 5 raised to the power 0.10 is 1.1746, giving 117.5 kg.
The other two follow the same weight and reps. O’Conner adds a flat 2.5% per rep: 1 + 0.025 × 5 is 1.125 — coincidentally identical to Brzycki’s multiplier at this exact rep count — so 112.5 kg again. Wathan’s exponential works out to 100 ÷ (48.8 + 53.8 × e^(−0.075 × 5)) × 100, or about 116.58 kg, which rounds to 116.6 kg.
The classic three land at 116.7, 112.5 and 117.5 kg. Their average is 115.6 kg, and the gap between the highest and the lowest is 5.0 kg — a little over 4% of the estimate. That gap is the honest precision of the whole exercise, and it is what this page headlines: O’Conner and Wathan are shown alongside the three for comparison, but neither joins that average. Anyone quoting an estimated max to the nearest kilogram is quoting a figure their method cannot support.
Now the training-load table, which runs from 95% down to 50% and is built from whichever formula Advanced options has selected — Epley by default, so untouched, it starts from 116.7 kg. Eighty percent of that is 93.4 kg, and Brzycki inverted (the table always uses Brzycki’s inversion for the reps column, whichever formula supplied the weight) says a typical lifter gets eight reps there: 37 − 36 × 0.80 = 8.2, rounded to 8. Open Advanced options and switch the formula picker to Brzycki instead and the table re-anchors on 112.5 kg — eighty percent of that is 90.0 kg, still eight reps by the same inversion, because the reps column only depends on the percentage. The headline stays 115.6 kg regardless of which formula the table is using; the advanced picker never touches it.
At 75% the inversion says ten reps, right at the line Mayhew et al. (2008) found these formulas reliable to. One row down, at 70%, it implies twelve — past ten, so that row is marked extrapolated rather than presented as equally solid. Below 70% the gap widens fast: 50% implies nineteen reps by Brzycki’s inversion and over fifty by Wathan’s, which is why the table stops trusting any of them past the ten-rep line rather than picking whichever inversion looks most reasonable.
Run the original set at twelve reps instead of five and the picture changes. Brzycki gives 144 kg, Epley 140 kg, Lombardi 128.2 kg, O’Conner 130 kg, Wathan 141.5 kg: the same lift, the same arithmetic, and now a 15.8 kg spread across the classic three. That is the cost of estimating far from a single, and because twelve reps is past the ten-rep accuracy line, this page also shows a visible note that the estimate is a rough guide rather than a precise one. It is the reason the sets people use for this are heavy triples and fives.
What these formulas are, and are not
None of the five is derived from physiology. Each is a curve that somebody fitted to rep-max data they had in front of them, and the data came from American collegiate strength and conditioning in the 1980s and early 1990s: young, trained lifters, overwhelmingly male, performing barbell lifts under supervision. Boyd Epley’s chart came out of the weight room he ran at Nebraska. Matt Brzycki’s appeared in a physical-education journal in 1993. Vincent Lombardi’s was published in a beginning weight training text in 1989. O’Connor, Simmons and O’Shea’s appeared in a 1989 weight training text of their own, and Dan Wathen’s was published as a book chapter on load assignment in 1994. None was validated on a large or diverse sample before it entered general use, and their durability owes more to being printed on gym walls than to the evidence behind them.
That history shows up in five practical ways.
- Accuracy degrades measurably above ten reps. Mayhew et al. (2008), comparing fourteen published prediction equations against measured 1RMs, found predictions held up best when reps-to-fatigue stayed at or below ten — a ceiling Brzycki’s own 1993 paper had already recommended. This calculator still accepts up to twelve, because a set that size is real information and the fits do not become meaningless the instant they pass ten, but it visibly flags the result and the training-load table’s rows once they do, rather than staying silent up to the harder cap described next.
- The input is capped at twelve entirely. Past that, the fits are extrapolating rather than interpolating, and Brzycki fails most visibly: its denominator is 37 − r, so it climbs toward infinity as reps approach 37 and goes negative past it. A set of twenty is measuring muscular endurance, a different quality from maximal strength, so this calculator refuses the input rather than returning something confident and wrong.
- The estimates diverge the further you are from a single. The classic three curves are anchored near the same point and spread apart as you move away from it — about 3% apart at one rep, 4% at five, 12% at twelve. The spread reported on this page is that divergence, and it is the closest thing to an error bar the method offers. O’Conner and Wathan follow a similar pattern without joining that figure.
- Rep capacity is individual. Two lifters with an identical true one-rep max can differ by four or five reps at 80% of it, because how many reps you get at a percentage depends on fibre composition, training history, and how much of your strength is neural rather than muscular. The training-load table is a population average dressed as a personal prescription.
- They say nothing about safety. An estimate is arithmetic on a number you already achieved. Whether you should walk up to that load is a question about your technique, your recovery, your equipment, and who is standing behind you — none of which appears anywhere in these formulas.
What they are good for is planning. If you need a percentage to load a training block against, and you would rather not test a true max to get one, a heavy set of three to five run through these formulas is a cheap and reasonable substitute — provided you treat the answer as approximate, which the spread on this page makes hard to forget.
What this calculator assumes
- The set was taken close to failure with clean technique. Reps left in reserve make the estimate too low; reps completed with form breakdown make it too high.
- The weight includes the bar. A 20 kg barbell with two 20 kg plates a side is 100 kg, not 80.
- The lift is a barbell movement with a stable path. Machines change the resistance curve and dumbbells change the stabilisation demand, so both shift the load–repetition relationship.
- Reps are whole and between 1 and 12. Outside that range the formulas are extrapolating, and Brzycki breaks down entirely as reps approach 37.
- No conversion happens between units. The formulas are ratios, so the unit toggle only decides which plausibility bounds apply.
- The training-load table’s rep column comes from Brzycki inverted regardless of which formula supplies the weight, and rows implying more than ten reps are marked as extrapolated because they sit past the accuracy line Mayhew et al. (2008) found for these equations.
- Advanced options lets you pick which of the five formulas anchors the training-load table; it defaults to Epley and never changes the headline average of Epley, Brzycki and Lombardi.
- O’Conner and Wathan are compared alongside the classic three but are not included in the headline average or the spread figure, which both remain three-formula numbers for backward compatibility with links shared before these two were added.
One-rep max FAQ
Which one-rep max formula is the most accurate?
None of them, reliably. Comparison studies find the ranking changes with the lift and with the population — Brzycki tends to read closer on the bench press, while Epley and the Mayhew family tend to hold up better on the squat and deadlift, and the differences are often smaller than the measurement noise. That is exactly why this calculator shows all three and reports how far apart they are instead of picking one and hiding the disagreement. If you want a single number, the average of the three is as defensible as any individual formula and has the advantage of sitting inside the range rather than at one edge of it.
Why can I only enter up to 12 reps?
Because past that the formulas stop describing strength. All three were fitted in the low rep range, and their error grows quickly outside it. Brzycki fails most visibly: it divides by 37 minus your rep count, so at 30 reps it returns an absurd figure, at 37 it divides by zero, and beyond 37 it returns a negative one-rep max. There is also a physiological reason the cap is not arbitrary — a set of 20 is largely testing muscular endurance, which is a different quality from maximal strength and is not recoverable from a load and a rep count. Refusing the input is more honest than returning a number that looks just like a good one.
Why does Epley give more than the weight I actually lifted for a single?
Because Epley multiplies your weight by 1 plus reps over 30, and at one rep that is 1.033 rather than 1. So 100 kg for a true single comes back as 103.3 kg. It is a known artefact of the curve fit, not a bug in this page, and it is left visible rather than patched over — Brzycki and Lombardi both collapse to exactly the weight you lifted at one rep, so seeing the three disagree at r = 1 tells you something real about how much shape each formula is imposing on the data. If you genuinely lifted a true one-rep max, you do not need an estimate: the weight on the bar is the answer.
Why do the estimates disagree more at higher reps?
Because they are three different curves through roughly the same low-rep data, and curves that agree near their anchor point diverge as you move away from it. At one rep the three sit within 3.3% of each other; at five reps they span about 5 kg on a 100 kg lift; at twelve reps the gap is nearer 16 kg. The spread this calculator reports is a rough confidence interval you can read directly — if the three formulas cannot agree within 15 kg, no single figure among them deserves to be treated as precise.
Should I actually test my one-rep max?
That is a question for you and someone who can watch you lift, not for a calculator. A true maximal attempt puts peak load on connective tissue and demands technique that holds up when it is hardest to hold up, and the risk is not the same for a lifter with five years of practice and a spotter as it is for someone eight weeks into training alone. The reason estimates like these exist at all is that a heavy set of three to five is a much cheaper way to get a usable number. This page will give you that number; it cannot tell you whether attempting the real thing is a good idea.
Does this work for dumbbells, machines, and bodyweight movements?
Less well, and for different reasons in each case. The formulas were built around barbell lifts, where load is continuous and the movement path is stable. Machines change the resistance curve, so the reps you get at a given percentage will not match the table. Dumbbells jump in fixed increments and demand more stabilisation, which compresses the useful rep range. Bodyweight movements have no adjustable load at all, so a rep count is not measuring the same thing. Use it for the barbell squat, bench, deadlift and press, and treat anything else as a rough starting point.
How should I use the percentage table?
As a starting point for loading, not as a contract. The load column is straightforward arithmetic — that percentage of the estimated max. The rep column is the load–repetition relationship inverted, and it says what a typical trained lifter should get at that percentage. Real people vary a lot around it: rep capacity at a given percentage depends on the lift, on training history, and on how much of your strength comes from neural efficiency versus muscle mass. If you consistently get four more reps than the table predicts, the table is not wrong about you so much as your estimated max is probably low.
What are the O'Conner and Wathan formulas?
Two more curve fits from the same era, added here so the comparison is not stuck at three. O'Conner (O'Connor, Simmons and O'Shea, 1989) is the plainest of the five: it adds a flat 2.5% of the working weight for every rep, so it is a straight line with no curvature at all. Wathan (Wathen, 1994) goes the other way — an exponential decay that, unlike Epley or O'Conner, does not grow without limit as reps climb; it settles toward a ceiling instead. Mayhew and colleagues' 2008 comparison of fourteen published equations found Wathan among the more accurate of the group. Both are shown next to Epley, Brzycki and Lombardi in the comparison list, but neither joins the three-formula average that drives the headline.
Can I choose which formula the training-load table uses?
Yes, in Advanced options. Epley is the default because it is the formula most training programs already quote by name, but the picker there covers all five. Changing it only moves the training-load table below — it re-anchors on whichever estimate you pick and the rows recompute from that figure. The headline "Estimated one-rep max" figure at the top of the page does not move; it stays the average of Epley, Brzycki and Lombardi no matter which formula the table is using, so the two numbers deliberately answer different questions.
Why does an accuracy note appear at 11 reps when the input still allows up to 12?
Because they are two different lines, marking two different claims. Twelve is the point past which these formulas stop describing strength at all — the input simply refuses a thirteenth rep, for the reasons in the question above. Ten is a tighter, evidence-based line: Mayhew et al. (2008), comparing fourteen published prediction equations against measured 1RMs, found the predictions held up best when reps-to-fatigue stayed at or below ten, a ceiling Brzycki's own 1993 paper recommended as well. Eleven and twelve reps are real sets and still produce a number here — the input cap did not move — but this page says plainly, both next to the reps field's result and in the training-load table, that the formulas are past the range the comparison literature actually checked, rather than staying silent about it until the hard cap at twelve.
Sources and review notes
- Epley, B. (1985). Poundage Chart. Boyd Epley Workout, Body Enterprises, Lincoln, Nebraska — the original source of the 1 + r ÷ 30 multiplier, from a collegiate weight room rather than a laboratory.
- Brzycki, M. (1993). “Strength testing: predicting a one-rep max from reps-to-fatigue.” Journal of Physical Education, Recreation & Dance 64(1), 88–90 — the 36 ÷ (37 − r) formula and the load–repetition table this page’s percentage column inverts.
- Lombardi, V. P. (1989). Beginning Weight Training: The Safe and Effective Way. Wm. C. Brown — the source of the r^0.10 power fit.
- O’Connor, B., Simmons, J., & O’Shea, P. (1989). Weight Training Today.St. Paul, MN: West Publishing, pp. 201–204 — the source of the 1 + 0.025r multiplier this page labels O’Conner, the spelling most gym-facing material uses for the same equation.
- Wathen, D. (1994). “Load assignment.” In T. R. Baechle (Ed.), Essentials of Strength Training and Conditioning(pp. 435–439). Champaign, IL: Human Kinetics — the source of the exponential-decay formula this page labels Wathan.
- Mayhew, J. L., Johnson, B. D., LaMonte, M. J., Lauber, D., & Kemmler, W. (2008). Accuracy of prediction equations for determining one repetition maximum bench press in women before and after resistance training. Journal of Strength and Conditioning Research, 22(5), 1570–1577 — reproduces the O’Conner and Wathan formulas verbatim alongside twelve others, and is the source of this page’s ten-reps-or-fewer accuracy guidance
- PubMed — the published literature comparing 1RM prediction equations across the bench press, squat and deadlift, which is where the “it depends on the lift” answer above comes from
- National Strength and Conditioning Association — official training-load chart, the published load–repetition table this page’s training-load table is checked against and tracks within a rep of everywhere the two overlap
- National Strength and Conditioning Association — the professional body whose textbooks publish the standard load–repetition percentage tables these figures are checked against
The arithmetic on this page is five multiplications and cannot go stale. The formulas themselves are curve fits three to four decades old, and nothing published since has replaced them so much as added to the pile — which is a reason to read the spread as seriously as the average.