The one number that isn’t one number
A one-rep max is the heaviest weight you can lift once, with good form, for a single repetition. Nobody wants to test it directly every week — a true max attempt is fatiguing and, near your limit, the riskiest set you will do. So lifters estimate it: lift a lighter weight for reps, run the numbers through a formula, and read off the max.
That is where the trouble starts. There is not one formula. There are several, all published, all respectable, all fitted to real lifting data — and they disagree. Run the same set through all six and you get six answers. A calculator that shows you one of them, without saying which, is not lying to you exactly. It is hiding the disagreement. That is worse, because a single confident number gets believed.
This guide shows the disagreement with real arithmetic, explains why it happens, and gives you the two fixes that actually work: use the range, and calibrate to yourself.
Six formulas, one set: the arithmetic
Take a single set — 100 kg for 5 reps — and run it through the six formulas our 1RM calculator uses:
| Formula | Equation | 100 kg × 5 |
|---|---|---|
| Epley | w × (1 + r / 30) | 116.7 kg |
| Brzycki | w × 36 / (37 − r) | 112.5 kg |
| Lombardi | w × r^0.1 | 117.5 kg |
| O’Conner | w × (1 + r / 40) | 112.5 kg |
| Mayhew | 100w / (52.2 + 41.9 × e^(−0.055r)) | 119.0 kg |
| Lander | 100w / (101.3 − 2.67123r) | 113.7 kg |
Same set. Answers from 112.5 to 119.0 kg — a 6.5 kg spread, about 6% of the median (115.2 kg). Every formula is “right” in the sense that each was fitted to real lifters’ data. They just fit different lifters, on different lifts, with different curve shapes.
Now push the reps up. The same 100 kg for 12 reps:
| Formula | 100 kg × 12 |
|---|---|
| Epley | 140.0 kg |
| Brzycki | 144.0 kg |
| Lombardi | 128.2 kg |
| O’Conner | 130.0 kg |
| Mayhew | 135.4 kg |
| Lander | 144.4 kg |
Now the spread is 128.2–144.4 kg — 16.2 kg, nearly 12% of the median (137.7 kg). The disagreement roughly doubles between 5 and 12 reps. At 10 reps it sits in between (125.0–134.1 kg, about 7%). You can reproduce every number in these tables with a pocket calculator; nothing here is a study result or a survey claim. It is what the formulas say.
That is the core finding, and it matches what lifters report in forums and groups: calculators behave up to about 5 reps, then drift. One lifter’s widely shared complaint was that a 15-rep set predicted a max he could not get within 35 kg of. Whether his exact numbers were typical does not matter — the mechanism is visible in the tables above. High-rep sets are where formulas diverge, so high-rep sets are where single-number calculators mislead most.
Why the formulas disagree
Three reasons, all ordinary.
They were fitted to different people. A formula fitted to trained powerlifters will not match a formula fitted to college students doing their first testing session. Rep-to-max relationships vary with training age, muscle fibre mix, technique, and which lift you are doing. A formula is a population average wearing a lab coat.
They assume different curve shapes. Epley is linear in reps. Lombardi is a power law. Mayhew is exponential. These curves agree in the low reps — they were all anchored near the same data there — and fan apart as reps rise, because nothing in the data forces them to stay together out there. Extrapolation is where curve shapes show their differences.
They ignore the lifter in front of them. Every formula takes two inputs: weight and reps. It does not know your leverages, your fatigue that day, or whether the set was a grinder or smooth. Two lifters can lift the same weight for the same reps and have genuinely different maxes. No two-input formula can capture that. The formulas are not broken; the expectation that one formula fits everyone is.
Even at one rep, the formulas do not fully agree: on a 100 × 1 set they return 100.0–108.9 kg. A formula is a fitted curve, not a law of physics — at r = 1 most of them do not return exactly the weight lifted, and that is normal for a fit. The honest way to read any estimate is as the centre of a range, not a point.
The fixes that work
1. Use the range, not the number
When six formulas span 112.5–119.0 kg, the useful information is the span. Plan around the middle, and know the edges. If your program calls for 85% of max, that is 95.6 kg off the low end and 101.2 kg off the high end — a 5.6 kg difference in the weight on the bar. Knowing that stops you from treating the estimate as exact. Our calculator’s Consensus mode reports the median plus the full spread for exactly this reason.
2. Calibrate to yourself
The second fix is better, and it is the one nobody had productized. A lifter in a forum thread once described doing it by hand: he logged his best sets at several rep ranges, took the middle of the estimates from his own sets, and predicted his max better than any standard calculator did. That makes sense — the inputs were his own sets, not a population average.
Our calculator’s My Median mode does the same thing without the spreadsheet: enter two or three sets you actually lifted, and it estimates a max from each set, then takes the median. Example: 100 kg × 5 gives an Epley estimate of 116.7 kg; 80 kg × 10 gives 106.7 kg; the median of the two is 111.7 kg. Two sets from different rep ranges anchor the estimate better than one set can, because they carry information about how your strength drops off with reps.
Calibration does not make the estimate exact. It makes it yours. That is the biggest available improvement, and it costs one extra set of typing.
3. Correct for effort with RPE
A formula assumes the set was close to failure. If it was not, the estimate overshoots. The RPE correction handles this: a set of 100 × 5 at RPE 8 counts as roughly 7 effective reps (5 + (10 − 8)), which Epley turns into 123.3 kg instead of the 116.7 kg a straight 5-rep reading would give. Wait — that is higher, not lower. Why? Because at RPE 8 you left two reps in the tank: you could have done 7 reps at that weight, and 7 reps predicts a higher max than 5. The correction is honest about what the set means. If you log RPE, use it — the RPE 1RM Calculator applies the correction for you.
How to use estimates without fooling yourself
- Stick to ≤5 reps for the inputs. The formulas agree most there (about 6% spread). Above 10 reps, treat any single number with suspicion.
- Compare like with like. Track progress with the same formula and the same mode every time. A “gain” that comes from switching formulas is not a gain.
- Use ranges for programming. Work from the middle of the range, and know the edges before you load the bar.
- Re-estimate after a training block. Estimates go stale as you adapt. A fresh hard set beats an old estimate.
- Never treat an estimate as a test. If the number matters — a competition opener, a max attempt — test it properly under the conditions that matter, with spotters and safety equipment. An estimate is a starting point for that test, not a substitute.
What this guide does not claim
No formula here diagnoses, treats, or prevents injury, and none of this is coaching advice. Strength training carries risk; a calculator cannot tell you what is safe for you to lift. The numbers are performance math — estimates with error bars — and the error bars are the point. See the Disclaimer.
Try it
Run your own last hard set through the One-Rep Max Calculator — Consensus mode shows you the six answers and the spread; My Median takes the median of the estimates from your sets. Then load the bar with the Plate Calculator, which solves the plates for whatever percentage you pick.
FAQs
Are 1RM calculators accurate?
Within limits. From a hard set of 1–5 reps, most formulas land within a few percent of a tested max for most lifters. From high-rep sets, the spread between formulas reaches ~12%, so any single number can be well off. The calculator is accurate in the same sense a weather forecast is: useful, honest about uncertainty, and wrong if you read it as a promise.
Why do the formulas disagree more at high reps?
Because they are different curve shapes fitted to different groups, and curves that agree at low reps fan apart when extrapolated. At 5 reps the six formulas span about 6%; at 12 reps, about 12%. The divergence is a property of the math, visible in the tables above — not a sign that one formula is “wrong” and the rest “right”.
Which formula is best?
No single winner. Epley is the most common default and a reasonable starting point. Brzycki runs low at higher reps; Mayhew tends to run high. The practical answer is to stop picking one: use the median of several (Consensus mode) and, better, calibrate to your own sets (My Median). The best estimate for you is the one built from your own sets.
Should I test my true 1RM instead of estimating?
If you need an exact number — a competition opener, a program built on true maxes — then yes, test it, with spotters, safety bars, and a proper warm-up. For week-to-week programming, estimates are enough, and they are far less fatiguing. Use estimates to plan; use tests to confirm, rarely.
How do I make an estimate more useful?
Three things, in order of impact: use a hard set of 1–5 reps as the input; enter two or three sets from different rep ranges so the estimate is calibrated to you; and read the result as a range, not a point. The calibration step is the one most lifters skip, and it is the one that helps most.
See also
- Deadlift 1RM Calculator — a deadlift estimate with dead-stop and touch-and-go notes.
- Bench Press 1RM Calculator — a bench estimate with paused and touch-and-go notes.
- Squat 1RM Calculator — a squat estimate with depth and bar-position notes.
- Overhead Press 1RM Calculator — a press estimate with strict and push-press notes.