Bench Press 1RM Explained: Why the Formulas Disagree and When to Trust Them

Four published formulas will turn the same bench press set into four different one-rep max estimates — close together at 5 reps, more than 30 kg apart at 15. Here is why the curves diverge, how to correct an estimate when your set wasn't truly to failure, and how to program off the number once you have it.

#health#strength#bench-press#1rm#weightlifting

What a 1RM formula actually claims

Every rep-max formula makes the same wager: the relationship between how much you can lift and how many times you can lift it is regular enough to run backwards. If 80 kg for 6 hard reps is your limit today, the formulas claim they can tell you what a single all-out rep would have been — without you ever attempting it. The bench press calculator runs four of the published versions side by side: Epley (1985), Brzycki (1993), Lombardi (1989) and O'Conner (1989), plus their average and a rep-max table derived from the result.

The calculator page already covers the basics — what a one-rep max is, why estimating beats testing for most people, and which rep ranges give trustworthy inputs. This guide goes into what the page doesn't: why four respectable formulas give four different answers, how far apart they drift and where, how to correct an estimate when your set wasn't truly to failure, and how to turn the number into working weights you can actually load on the bar.

Why four formulas give four answers

The formulas disagree because they are three different curve shapes fitted to different groups of lifters. Epley and O'Conner are straight lines: each rep adds a fixed fraction of the bar weight to the estimate — 1/30th per rep for Epley, 1/40th for O'Conner, which is why O'Conner always reads lowest or joint-lowest. Brzycki is a hyperbola, 36/(37 − r): it behaves like the linear formulas at low reps but bends upward steeply as reps climb, heading for infinity at 37 reps, where the arithmetic breaks down entirely. Lombardi is a power curve, r0.1, which rises quickly over the first few reps and then almost flattens.

In the range the formulas were built for — roughly 2 to 10 reps — the shapes barely differ, and neither do the answers. Feed the calculator 80 kg × 6 and the four estimates span just 4 kg: 92.0 to 96.0. There is even a neat crossover point: at exactly 10 reps, Epley and Brzycki agree to the last decimal — both return 4/3 of the bar weight, so 60 kg × 10 gives 80.0 kg from each. Below 10 reps Brzycki reads slightly lower than Epley; above 10, slightly higher.

Beyond 10 reps the shapes fly apart, and so do the answers. A 15-rep set of 100 kg returns 131.1 kg from Lombardi, 137.5 kg from O'Conner, 150 kg from Epley and 163.6 kg from Brzycki — a spread of more than 32 kg from one input. None of those is a prediction worth programming from; the spread itself is the message. High-rep sets measure muscular endurance as much as strength, and lifters vary enormously in endurance at the same max. That is why research on prediction accuracy — for example Reynolds and colleagues (2006) — consistently finds low-rep sets predict best, and why the calculator flags anything above 10 reps as a rough figure only.

Worked example: 80 kg for 6 reps

Say you bench 80 kg (176 lb) for 6 full reps and the seventh isn't happening. Here is what each formula does with that set:

  • Epley: 80 × (1 + 6/30) = 80 × 1.2 = 96.0 kg
  • Brzycki: 80 × 36 ÷ (37 − 6) = 2,880 ÷ 31 = 92.9 kg
  • Lombardi: 80 × 60.1 = 80 × 1.1962 = 95.7 kg
  • O'Conner: 80 × (1 + 0.025 × 6) = 80 × 1.15 = 92.0 kg

The average of the four is 94.2 kg, and the honest reading is “my max is somewhere in the low-to-mid 90s.” The bench press calculator then inverts the headline Epley estimate into a rep-max table: from a 96 kg max, the heaviest you can likely handle is 90.0 kg for a double, 87.3 kg for a triple, 82.3 kg for 5 reps, 75.8 kg for 8 and 72.0 kg for 10. Notice the self-consistency check hiding in that table: invert 96 kg back to 6 reps and you recover exactly the 80 kg you started with. The same arithmetic works identically in pounds — 185 lb × 5 estimates a 216 lb max — because every formula scales linearly with the weight; only the rep count bends the curve.

From estimate to training weights: percentages and the training max

The practical value of a 1RM is that programmes prescribe loads as percentages of it. The Epley inversion maps reps to percentages directly: 100 ÷ (1 + r/30) gives roughly 93.8% of max for a double, 90.9% for a triple, 85.7% for 5 reps, 78.9% for 8, 75% for 10 and 71.4% for 12. If your programme says “5 × 5 at 80%,” the table turns that into a bar weight in one step — from the 96 kg estimate above, 80% is 76.8 kg, call it 77.5 kg with standard plates.

Many popular programmes deliberately don't use your true max at all. They calculate from a training max — typically 90% of your tested or estimated 1RM — precisely because daily readiness fluctuates and estimated maxes carry a few percent of error. Off the 96 kg estimate, a 90% training max is 86.4 kg, and every percentage in the programme keys off that. Building this margin in is a feature, not caution for its own sake: it means a slightly optimistic estimate doesn't poison every session that follows. If you use a training max, feed the programme's percentages from it consistently — don't mix true-max and training-max percentages in the same cycle.

What if the set wasn't to failure? The reps-in-reserve correction

Every formula assumes the input set was a genuine limit — that one more clean rep did not exist. Training sets usually aren't like that, and stopping short is often the right call. The fix is simple: add the reps you had left in the tank before running the numbers. Sports scientists formalise this as reps in reserve (RIR), the backbone of the resistance-training RPE scale validated by Zourdos and colleagues (2016) — RPE 8 means two reps in reserve, RPE 9 means one.

The correction matters more than most lifters expect. Take the 80 kg × 6 set above, but suppose it was RPE 8 — two reps in reserve. Your true capability that day was 80 kg × 8, and Epley on the corrected set gives 101.3 kg, not 96.0 — a difference of more than 5 kg from two unlogged reps. Estimating your own RIR accurately is itself a skill (Helms and colleagues (2016) offer a practical guide), and lifters new to it tend to underestimate how much they have left. If you aren't sure, take a set genuinely close to failure once in a while specifically to calibrate — or accept that your estimate is a floor, not a ceiling.

Factors that decide how accurate your estimate is

Rep count of the input set

The single biggest factor. Formulas were fitted mainly on sets of 10 reps and fewer, and bench press studies find predictions from sets of about 6 reps or fewer land within a few percent of tested maxes. By 15 reps the four formulas can't even agree with each other, let alone with the bar.

Proximity to failure

An estimate from a set with three quiet reps in reserve understates you by design. Either push the calibration set hard or apply the RIR correction above — but never both, or you double-count.

Training experience

Maximal lifting is a skill. A newer lifter's tested max often falls short of the formula estimate not because the muscles aren't there but because bracing, bar path and confidence degrade under loads they've never felt. Experienced lifters convert estimated strength into tested strength more completely.

Your endurance profile

Two lifters with identical 100 kg maxes might manage 4 and 9 reps at 85 kg. Rep-based formulas can't see that difference — they assume an average endurance profile. If you know you're an outlier at high reps, trust estimates from your low-rep sets and treat the rest as noise.

Day-to-day readiness

Sleep, stress, bodyweight changes and accumulated fatigue swing daily strength by several percent. An estimate is a snapshot of the day it was taken, which is one more argument for programming off a discounted training max.

How to raise your bench press 1RM

  • Add load or reps gradually, and re-estimate rather than re-test. Log a hard set every week or two, run it through the calculator, and watch the trend. The estimate creeping from 96 to 101 kg over a cycle is progress you can see without ever staging a max attempt.
  • Bench more often. Pressing twice or three times a week at moderate intensities generally beats one heroic session — frequency is practice, and maximal pressing is a skill.
  • Tighten the technique that only shows up near limit loads. Shoulder blades pinned, feet planted for leg drive, a consistent touch point and a grip width you can reproduce. Technique leaks cost more at 95% than they ever do at 70%.
  • Train the supporting cast. Close-grip bench, dips and overhead pressing for the triceps and shoulders; rowing for the upper back that stabilises the press.
  • Feed and recover for it. Strength gains are built between sessions. Adequate daily protein — the protein calculator gives a target from your bodyweight and training load — and sufficient total calories matter; your BMR plus activity expenditure is the baseline to eat around.

Common mistakes

Feeding it a high-rep set. A 15-rep burnout set produces estimates spanning 30+ kg depending on the formula — the maths is telling you it doesn't know. Use a set of 10 or fewer, ideally 3–8.

Counting reps that wouldn't count.Bounced, half-locked or increasingly shallow reps inflate the rep count and the estimate with it. Count the reps a strict judge would.

Getting the bar weight wrong. The estimate is only as good as the weight you enter. A standard bar is 20 kg (45 lb) — two 20 kg plates make 60 kg total, not 40. Count every plate and the bar, and enter total bar weight.

Jumping straight to the predicted single. An estimate is a planning number, not a warranty. If you do want to test a true max, work up through progressively heavier singles with full rest, with a spotter or safety bars, and treat the estimate as the ceiling for the day's final attempt — not the opener.

When a calculator isn't enough

A formula can't watch your bar path. If you're preparing for a powerlifting meet, coming back from a shoulder or pec injury, or you feel pain (rather than effort) when pressing, get eyes on your lifting — a qualified coach or, for pain, a medical professional. Estimation tools sit safely inside a sensible programme; they don't replace one, and nothing here is individual training or medical advice.

Frequently asked questions

The FAQ on the bench press calculator page covers the definition of a 1RM, why estimating beats testing, formula accuracy, input rep ranges and training percentages. The questions below cover what comes up once you start using the number: retesting cadence, not-to-failure sets, training maxes and why the bar sometimes disagrees with the maths. For the wider training picture, see the protein guide and the BMR guide.

Frequently asked questions

How often should I re-estimate my bench press 1RM?

Log a hard working set every week or two and run it through the calculator — the trend across a training cycle is the useful signal, not any single reading. Re-estimating is free, unlike max testing, so there is no reason to let the number your programme runs on go stale for months. Most lifters update their training max between cycles, every 4–6 weeks, using the best recent estimate.

My set wasn't taken to failure — can I still use it?

Yes, with a correction: add the reps you honestly had left before entering the set. A set of 80 kg × 6 with two reps in reserve reflects a true capability of 80 kg × 8, which raises the Epley estimate from 96.0 kg to 101.3 kg — more than 5 kg from two unlogged reps. If you can't judge your reps in reserve confidently yet, treat the uncorrected estimate as a floor rather than your actual max.

Why is my tested max lower than the calculator predicted?

Usually one of three reasons. Maximal lifting is a skill — bracing, bar path and confidence degrade under loads you have never handled, especially for newer lifters, so estimated strength doesn't fully convert. Daily readiness also swings strength by several percent, so a max tested on a flat day undershoots an estimate made on a good one. And if the input set was high-rep, the estimate itself was soft: predictions from sets over 10 reps run optimistic for lifters with good muscular endurance.

Should I program from my true 1RM or a training max?

Many well-known programmes deliberately use a training max — commonly 90% of your tested or estimated 1RM — and calculate every percentage from that. The discount absorbs estimation error and day-to-day fluctuation, so a slightly optimistic max doesn't derail whole weeks of training. Whichever you use, be consistent: mixing true-max percentages and training-max percentages in one cycle defeats the point of either.

Does the bar count in the weight I enter?

Yes — enter the total weight lifted, which is the bar plus every plate. A standard Olympic bar is 20 kg (45 lb), so a bar loaded with one 20 kg plate per side is 60 kg total, not 40. Miscounting the bar is one of the most common ways an otherwise careful estimate ends up 20 kg wrong, and collars add a little more on calibrated setups.

Can beginners use a 1RM calculator?

Yes — arguably beginners benefit most, because a true max attempt is exactly what a new lifter shouldn't be doing while technique is still settling. Estimate from a set of 5–8 solid reps instead and use the percentages for programming. Expect the estimate to move quickly: novice strength and skill improve fast, so re-estimate often and don't be surprised if a tested max lags the prediction until maximal technique catches up.

Is the average of the four formulas better than one formula alone?

Within the reliable 2–10 rep range the four estimates sit close together, so the average mostly just smooths a small spread — from 80 kg × 6 the estimates run 92.0 to 96.0 kg and the average is 94.2 kg. No study crowns a single formula for every lifter, so the average is a reasonable default, and the spread itself is useful information: when the four numbers diverge widely, your input set was too long to trust regardless of which formula you pick.

Informational only. Not personalised financial, legal, or tax advice.