One-Rep Max Calculator

Your one-rep max is estimated by averaging Epley (weight × (1 + reps ÷ 30)) and Brzycki (weight × 36 ÷ (37 − reps)) — two formulas that return the identical answer at exactly 10 reps, stay within 3.8% of each other at every whole rep count below it, and are 27.1% apart by 20 reps.

How the 1RM estimate works

A one-rep max is the heaviest weight you can move for a single clean repetition, and it's the backbone of most strength programs — sets are usually written as a percentage of it. Testing a true max is taxing and, for beginners, risky, so this calculator estimates it from a submaximal set instead: you enter a weight and the number of reps you completed with it.

The tool runs two widely used equations and shows you the average of the two. That single number is easy to read, and it hides something worth knowing, because how much the two equations agree depends entirely on the rep count you fed them. The rest of this page works out where the average is a fair summary and where it is papering over a real split. For the history of the two formulas and the research on how well either predicts a tested max, see our guide to calculating your one-rep max.

The two formulas agree at exactly 10 reps and at no other rep count

This is not an approximation. Set the two equations equal to each other and the lifted weight cancels straight out, leaving a statement about reps alone: (30 + r) × (37 − r) = 1080, which rearranges to r² − 7r − 30 = 0 and factors as (r − 10)(r + 3) = 0. The roots are 10 and −3. A set of −3 reps is not a set, so there is exactly one rep count at which Epley and Brzycki return the same number, and it is ten.

Away from that crossing they diverge, and they change places as they pass through it. Below 10 reps Epley returns the higher estimate; above 10 reps Brzycki does. Brzycki is the more conservative of the two only underneath the crossing; above ten reps it is the higher of the pair, and the table below gives the size of the swing.

The size of the gap is a property of the rep count and nothing else. Both formulas are the lifted weight multiplied by a factor that depends only on reps, so the weight cancels out of the comparison as well, leaving 1080 ÷ ((37 − r) × (30 + r)) − 1. At five reps that is 3.6% whether the bar holds 60 kg or 225 kg. Only the kilograms between the two answers scale with the load. The percentage does not, and neither does it change if you switch the unit selector to pounds.

How far apart the two formulas get

Both formulas evaluated on a 100 kg set and rounded to one decimal place, as the calculator prints them. Gap is Brzycki minus Epley as a share of Epley, so a negative figure means Epley is the higher of the two. The percentages in that last column are identical for any lifted weight and either unit, because the weight cancels; only the kilogram columns change.

Two things fall out of that last column. The first is that underneath the crossing the argument is small: across every whole rep count from 1 to 9 the two formulas stay within 3.8% of each other, and the widest of those gaps sits at 3 and 4 reps, where both come out to 3.74%.

The second is that the split is not symmetrical about the crossing, and it does not blow up the moment you pass ten. At 11 reps the two are 1.3% apart and at 12 reps 2.9% — still inside the range you get below the crossing. The real deterioration starts at 13 reps (4.7%) and accelerates from there: 9.1% at 15 reps, 18.4% at 18, and 27.1% at 20, which is about seven times the worst disagreement found anywhere below ten. The cause is structural rather than mysterious. Brzycki's denominator is 37 − reps, so it shrinks toward zero as the rep count climbs, and a shrinking denominator makes the estimate rise ever faster. Epley's multiplier just adds a flat 1/30 of the weight per rep, forever.

That matters because the tool shows you one number. At 20 reps on a 100 kg set, the 189.2 kg it prints sits 13.5% above what Epley says and 10.6% below what Brzycki says. It is not a compromise between two close readings. It is the midpoint of a 45 kg argument, and the averaging is what makes that argument invisible.

Enter one rep and the estimate still comes back heavier than the bar

Feed the calculator a genuine single — a weight you lifted once and could not have lifted twice — and it does not hand the weight back to you. It adds 1.7%. A 100 kg single is reported as 101.7 kg.

The reason is in the two multipliers. Brzycki's denominator at one rep is 37 − 1 = 36, so the formula is 36 ÷ 36 = 1 and it returns exactly what you entered; it is right by construction at that rep count. Epley's multiplier is 1 + 1/30, which adds 3.3% on top. Averaging the two keeps half of that overshoot. The amount is small, but it illustrates what averaging does in general: whichever formula is misbehaving at a given rep count, half of that misbehaviour survives into the number you see. The practical point is simpler than the arithmetic — if you tested a true single, that single is your max, and you do not need this calculator for it.

The one thing neither formula can see

Both equations take exactly two inputs: the weight and the rep count. Neither has any way to know how close the set was to failure. Two people can log an identical set and be in entirely different places when they rack the bar, and the formula sees no difference at all.

Two lifters both put 100 kg up for 5 and both get 114.6 kg back. Say the first was genuinely finished — a sixth rep was not there. Say the second racked it with three more in him. The set that would actually have measured the second lifter is 100 kg for 8, and this calculator returns 125.4 kg for that: 10.8 kg, or 9.4%, higher. Nothing in the input changed in a way the arithmetic could detect.

That is also why the standard advice — a hard set with about one rep left in the tank — is a deliberate trade rather than a free choice. Running the same arithmetic, one rep left in reserve on a five-rep set costs about 3.0% on the estimate; three reps in reserve costs the 9.4% above. You are buying a safer set at the price of a lower number. That is usually the right trade, but it is worth knowing which way it leans: reps left in reserve bias the figure downward every time, never upward, because the formula rises with every rep you feed it. It is a one-directional bias, not a coin flip.

Percentage of 1RM by reps

The third column inverts the averaged formula this tool ships: it is 100 divided by the multiplier the calculator applies at that rep count, so it is what the tool itself implies the lifted weight was as a share of the estimated max. Six of the eight rows match the conventional chart to within half a percentage point. The two end rows do not — the 100% row comes out at 98.4% and the 65% row at 63.8%. The first two columns are the conventional load-to-reps chart, which is approximate and varies between individuals. The result panel prints 95% down to 60% in five-point steps.

Agreement is not accuracy

It is tempting to read the crossing at ten reps as a sweet spot — the place where two independent methods confirm one another. It is not that. The crossing is what happens when two multipliers published eight years apart happen to intersect, and it says nothing at all about whether four-thirds of your ten-rep weight is really your max. Two formulas can agree perfectly and both be wrong by the same amount.

What the arithmetic on this page does license is narrower, and more useful. It tells you where the single number on screen is an honest summary of its own two ingredients, and that is at every whole rep count from 1 to 12, where they stay within 3.8% of each other. Past that the number is a summary of a disagreement. Whether a low-rep estimate is closer to a tested max is a separate question that no amount of algebra settles, and the research on it is set out in our guide to calculating your one-rep max.

Getting a useful number

Put the two together and the working rule is short. Use a heavy set you finished with good form, in the 2-to-10-rep range, with about one rep left in the tank — accepting the roughly 3% that reserve costs the estimate, for the reason set out above. That window sits entirely inside the zone where the calculator's two ingredients stay within 3.8% of each other; on the separate question of which rep counts predict a tested max most closely, our guide argues for the lower end of it, a set of about two to five reps. Treat anything estimated from a set of thirteen reps or more as a bracket rather than a figure — that is where the two ingredient formulas stop agreeing — and re-estimate from the same lift and a similar rep count each time so your comparisons are fair.

Use the percentages as a guide, not gospel. Strength blocks typically live at 80–95% of 1RM, muscle-building work around 65–80%, and lighter technique or conditioning work below that. Bar speed, sleep and stress all shift how a given load feels on the day, so adjust as needed. Strength is built between sessions as much as during them: our protein intake calculator sets a daily protein target, the TDEE calculator estimates the total food that supports recovery, and the macro calculator splits that total across protein, carbs and fat. This calculator tracks what you can lift; for the size side of training, our guide to how long it takes to build muscle covers what the research shows about timelines, the lean body mass calculator estimates how much of your weight is not fat, and the FFMI calculator puts that lean mass in context for your height.

Formulas as shipped in this calculator: Epley (1985), 1RM = weight × (1 + reps ÷ 30), and Brzycki (1993), 1RM = weight × 36 ÷ (37 − reps), averaged. Every figure on this page is computed from those two equations. Estimates are most reliable at 2–10 reps. Educational only, not medical advice — near-maximal lifting carries injury risk, so check with a clinician first if you have a heart condition, uncontrolled blood pressure, a hernia or a relevant injury history.