How to Calculate Your 1RM Without Maxing Out.

Every strength program asks for the same number. Your one-rep max. Plug it in, and the program gives you percentages. Work sets at 75%, top sets at 85%, openers at 92%. Without that number, you cannot use the program.

But getting it the honest way means loading a bar heavier than you have ever lifted and hoping today is the day. Maxing burns recovery, risks form breakdown at the worst possible moment, and produces a number only as good as the eight hours of sleep you got the night before. So everyone reaches for a calculator instead.

The calculator gives you a single confident number. Type in your last set of five at 225, get back 263 lb. Easy. Programmable. Reassuring.

It is also informed guesswork dressed up as math.

Five formulas. Four chart fossils..

RepRaptor's 1RM calc lets you pick from five formulas: Epley, Brzycki, Lombardi, O'Conner, and Wathan. Most calculators offer a similar set. What none of them tell you is where the formulas came from.

Boyd Epley was the strength coach at the University of Nebraska in the 1980s. In 1985 he published a manual called the Boyd Epley Workout for his football players. On page 86 there is a poundage chart, a lookup table that maps reps to estimated maxes. Years later someone algebraicized the chart into 1RM = w × (1 + r/30). That equation. The one in every calculator on the internet. It is a curve fit to a chart that one coach drew up for his football team. There is no peer-reviewed derivation. The constant 1/30 is a trial-and-error guess at strength drop-off per rep.

Lombardi (1989) is worse. Asked how he settled on the formula, the author admitted in writing that it was "a lot of guesswork and a bit of intuition… tinkering with the equation when I saw the data did not look right." O'Conner (1989) is a textbook artifact. Wathan (1994) is a chart from a chapter Doug Wathan contributed to the NSCA Essentials textbook. Lander (1985), which most calcs include in some form, is a chart from the NSCA Journal that got an equation later.

Brzycki (1993) is closer to research, but only just. Matt Brzycki extrapolated his formula from a graph in a 1981 textbook by Sale and MacDougall. The graph itself was drawn from unpublished observations by Anderson and Haring. The 1993 paper is a practitioner article in JOPERD with no original sample, no statistics, no validation. Brzycki is a curve drawn through a curve drawn through someone's notes.

Of the formulas most calculators offer, exactly one comes from a primary-data study. Mayhew and colleagues (1992) tested 184 men and 251 women at Northeast Missouri State on bench press, fit an exponential curve, and published it. It is the most defensible formula in the literature. RepRaptor's calc does not offer Mayhew. Almost no free calc on the internet does.

The chart above shows what happens when you feed five formulas the same numbers. At one rep they agree. At five, they spread across roughly 11 lb on a 225 set. At ten, the spread doubles to about 22 lb. At fifteen, Brzycki's denominator approaches zero and the math collapses, which is why the calc refuses to return a Brzycki number above 15 reps.

What the evidence actually says..

The canonical comparison study is LeSuer et al. (1997). They tested 67 college students on bench, squat, and deadlift, then compared seven formulas against the measured 1RM. Across all three lifts, the correlation between predicted and actual was above 0.95. High, but not as high as the rounding in your favorite calculator implies. Mayhew, Wathan, and Epley had the lowest absolute errors. Estimates were close on bench and squat. On deadlift, the errors were roughly twice as large.

The more interesting finding came nine years later. Reynolds, Gordon, and Robergs (2006) tested 70 subjects across 1, 5, 10, and 20RM on bench press and leg press. They found that estimation accuracy got worse as the rep-out set got longer. Five reps was the most accurate single estimator. Ten was worse. Twenty was much worse. This directly contradicts what every calculator implicitly tells you. More reps does not mean more data. More reps means more conditioning bleeds into the rep-out set, and conditioning is not strength.

DiStasio (2014) tested Division I college football players on back squat. He found the Epley equation significantly overestimated 1RM from 5-rep sets, but matched the true 1RM closely from 3-rep sets. The takeaway is uncomfortable. As you get stronger, the formulas need fewer reps to stay accurate. The "do up to 10 reps" boilerplate on most calculators is wrong for advanced lifters.

Nuzzo and colleagues (2024) ran the largest meta-analysis on this question to date. 952 reps-to-failure tests across 7,289 individuals from 269 studies. They looked at what moderates the reps-to-percent-1RM relationship. The result that should rattle every calculator: sex, age, and training status had small effects. The biggest moderator was exercise type. A bench-derived formula does not generalize cleanly to leg press, much less to deadlift or overhead press. The single universal calculator is structurally wrong.

For deadlift specifically, Lake et al. (2017) found that velocity-based estimation systematically underestimated the actual 1RM with effect sizes of d = 1.03 to 1.75. Meaningfully wrong, every time. The deadlift starts from a dead stop, has no stretch-shortening cycle, and ends with a more abrupt failure mode than the lifts the formulas were built on. None of the popular formulas handle it well.

Honest summary: at 3 to 6 reps on bench or squat, in a lifter who is trained but not elite, the better formulas land within about 5% of measured 1RM. Outside that window (deadlift, older or female populations, above 10 reps), expect 10 to 20% error, and do not bet anything important on the exact number.

So pick a formula. Here is how..

If you are going to use a 1RM calc (and most lifters will), here is what the evidence supports.

Do your rep-out at 3 to 6 reps. Five is the sweet spot. Below three, every formula returns roughly the same number, so it does not matter which you pick. Above six, the spread between formulas grows and the error compounds. Above ten, the result is closer to a guess than an estimate.

For main compound lifts in moderate rep ranges, pick Wathan or Epley. Both are among the best-performing formulas in the validation studies. Wathan's exponential shape handles the load-rep curve more honestly than Epley's linear form. Epley is fine and slightly conservative at low reps.

Avoid Brzycki above six reps. It is mathematically conservative at low reps but its denominator approaches zero as reps approach 37, which makes the curve unstable above moderate rep counts. The calc actively refuses Brzycki above 15 reps and warns above 10. Those limits are not arbitrary.

For deadlift, treat any calculated number with suspicion. If you genuinely need to know your deadlift 1RM, true-test it with a proper warm-up protocol, or use it for relative comparison between training blocks only. The formulas were derived from bench data and they show it.

The best feature of the RepRaptor calc is that it shows you the formulas disagree. Pick Epley and you see one number. Wathan gives you a different one. Lombardi spits out a third. That disagreement is information. It tells you the answer is a band, not a point. Treat your "1RM" as a range. Roughly plus or minus 5% on bench and squat at 5 reps, plus or minus 10% on deadlift. Program off the middle of that range, expect day-to-day fluctuation, do not bet a meet attempt on the exact number.

The deeper question..

Everything above assumes you need a fixed 1RM number to plug into your program. Maybe you do not.

Daily 1RM fluctuates. González-Badillo and Sánchez-Medina (2010) showed that a prescribed percentage of 1RM represents meaningfully different actual relative intensities on different days, depending on neuromuscular readiness, sleep, food, stress. The "85% of 1RM" you prescribe on Monday might feel like 80% if you slept well, and 90% if you did not. Anchoring all your training to a fixed number from a max test four weeks ago is anchoring to a snapshot that has already drifted.

The alternative comes from Mike Tuchscherer's Reactive Training Systems and was popularized by Eric Helms and Greg Nuckols at Stronger by Science. Anchor the prescription to perceived effort instead of to a fixed number. Instead of "5 reps at 78% of 1RM," prescribe "5 reps at RPE 8" (a set you finish with about two reps left in the tank). The weight that produces that effort today is, by definition, today's 78%. No four-week-old test number required.

The translation chart comes from Mike Tuchscherer's Reactive Training Systems. RepRaptor's RPE calc uses it directly.

To read the chart, cross a rep count with an RPE rating to get the percentage of today's 1RM that lift represents. A set of 5 at RPE 8 is 78.2% of today's max. A single at RPE 10 is 100% by definition. A triple at RPE 9 is 88.1%.

The advantage is not the math. You stop trying to outguess yourself, which is what actually matters. You do not need to remember a 1RM, recalculate when it changes, or feel guilty when "your 85%" suddenly feels heavier. You walk in, warm up, lift, rate the effort, and the math gives you a percentage of today.

There is a real catch. RPE accuracy. Zourdos et al. (2016) found that experienced lifters rate RPE reasonably well, but novices systematically under-report effort. A new lifter rating their set as RPE 8 is often actually at RPE 9 or 10. If you cannot reliably tell the difference between a set with one rep in reserve and a set with three, RPE-based programming will not work for you yet. Spend a month doing AMRAPs and getting calibrated before you switch.

When you actually should test..

Estimation is the right default. But there are situations where a true 1RM test is the correct call.

Meet prep is the obvious one. If you are competing in powerlifting or weightlifting, you need to know what you can lift on the platform under judging conditions. A peaking block typically includes at least one heavy single at RPE 9 or 10 within two or three weeks of the meet, plus an opener simulation around 92% of projected.

A new program with a coach who runs supervised testing is another. The NSCA protocol uses a brief warm-up, a 3-to-5 rep set at an estimated weight, then progressively heavier singles with 2 to 4 minutes rest between attempts. A true 1RM should be reached within three to five attempts. With proper warm-up and a competent spot, the injury risk is low.

Returning from a layoff is the trickier case. If you have been out for a month or more, your old 1RM is wrong. Test with a 3RM or 5RM and back-calculate. Better information per unit of risk than a true single on detrained tissue.

Do not test if you are a true beginner with less than 8 to 12 weeks of consistent lifting. Form is not consolidated, motor learning still dominates, and the number will change week to week regardless. Do not test if you have any of the contraindications the clinical guidelines list (uncontrolled hypertension, recent hernia or aneurysm, retinal detachment, recent joint surgery). Do not test as a flex. Test because you have a programming reason that estimation cannot solve.

Bottom line..

The number on a 1RM calculator is informed guesswork. The good formulas land within 5% on bench and squat at moderate reps in trained lifters. The bad ones, which is most of them, are charts that someone fit an equation to forty years after the fact. The deadlift formulas fail systematically. The "more reps means a better estimate" advice is wrong. So is the "one universal formula" framing.

Pick a low rep count, pick a defensible formula, treat the output as a range, and program off the middle. Or skip the fixed-1RM model entirely and anchor to today's effort with RPE. Both work. Neither lets you avoid actually putting in the work.

The bar does not care what your calculator said.