No one rep max formula wins every validation study, and the number of reps in your test set matters more than the choice between Epley and Brzycki. Feed the same 100 kg set into the seven common one rep max formulas and, between 4 and 10 reps, their answers stay within 5.6% to 7.8% of each other. At 15 reps they are 24.8% apart. Epley and Brzycki give the identical answer at exactly 10 reps.
General fitness information, not medical advice. The formulas are the seven repetitions-to-failure equations tabulated in a 2021 Scientific Reports paper that reproduces LeSuer and colleagues' 1997 comparison; every number in the grids below was computed by CalculatorVillage on October 30, 2026 from a hypothetical set of 100 kg. Study findings are stated only as far as each study's PubMed abstract states them, checked October 30, 2026.
Seven equations, one set of inputs
Each of these formulas takes two inputs: the weight on the bar and the number of reps you completed before you could not do another clean one. They differ in the shape of the curve they assume between those two things. Two are straight lines, two are hyperbolas, one is a power curve and two are exponential curves that level off.
| Formula (year as usually cited) | Equation used here | Shape | What it returns for a single rep of 100 kg |
|---|---|---|---|
| Epley (1985) | weight × (1 + reps / 30) | Straight line | 103.3 kg |
| Brzycki (1993) | weight × 36 / (37 − reps) | Hyperbola | 100.0 kg |
| Lander (1985) | 100 × weight / (101.3 − 2.67123 × reps) | Hyperbola | 101.4 kg |
| Lombardi (1989) | weight × reps^0.10 | Power curve | 100.0 kg |
| Mayhew et al. (1992) | 100 × weight / (52.2 + 41.9 × e^(−0.055 × reps)) | Exponential, levels off | 108.9 kg |
| O'Conner et al. (1989) | weight × (1 + 0.025 × reps) | Straight line | 102.5 kg |
| Wathan (1994) | 100 × weight / (48.8 + 53.8 × e^(−0.075 × reps)) | Exponential, levels off | 101.3 kg |
A few notes on the forms, because the same formula turns up written several ways:
- Epley is often printed as weight × (1 + 0.0333 × reps). That constant is 1/30 rounded, and the two versions differ by less than 0.1% even at 15 reps (1.4995 against 1.5). The grids here use reps / 30, which is also what the CalculatorVillage calculator uses. The Scientific Reports table prints Epley with an extra "1 ÷" in front of the bracket, which would make every estimate smaller than the weight lifted, so it reads as a typesetting slip; this page keeps that table's 0.0333 constant and bracket and drops the division.
- Brzycki appears as weight × 36 / (37 − reps), as weight / (1.0278 − 0.0278 × reps), and in the Scientific Reports table as 100 × weight / (102.78 − 2.78 × reps). At 5 reps all three give a multiplier of 1.125. They are the same equation with the constants rearranged.
- Wathan is spelled Wathen in many papers, including the PubMed abstracts cited below. O'Conner is spelled O'Connor in some. The equations are the same.
- The years come from the reference list of an open-access PeerJ paper that uses all seven (see Sources). That list also shows where they first appeared: several in a textbook, a coaching book or a professional association journal rather than in a validation study, which is one reason so much later work has gone into testing them.
The last column matters more than it looks. A one-rep max formula should, in principle, return the weight you lifted when you lifted it once. Only Brzycki and Lombardi do. Mayhew adds almost 9%. That does not make Mayhew a bad formula at the rep counts it was built for, but it shows these equations are curve fits over a working range, not laws. The calculator on this site sidesteps the problem by returning the lifted weight unchanged whenever you enter one rep.
The grid: 100 kg lifted for 1 to 15 reps
Here is every formula run on the same hypothetical set. Read across a row to see how far the formulas disagree about one performance. The last three columns show the lowest estimate, the highest, and the gap between them as a percentage of the lowest.
| Reps at 100 kg | Epley | Brzycki | Lander | Lombardi | Mayhew | O'Conner | Wathan | Low | High | Spread |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 103.3 | 100.0 | 101.4 | 100.0 | 108.9 | 102.5 | 101.3 | 100.0 | 108.9 | 8.9% |
| 2 | 106.7 | 102.9 | 104.2 | 107.2 | 111.4 | 105.0 | 105.1 | 102.9 | 111.4 | 8.3% |
| 3 | 110.0 | 105.9 | 107.2 | 111.6 | 114.0 | 107.5 | 109.0 | 105.9 | 114.0 | 7.7% |
| 4 | 113.3 | 109.1 | 110.4 | 114.9 | 116.5 | 110.0 | 112.8 | 109.1 | 116.5 | 6.8% |
| 5 | 116.7 | 112.5 | 113.7 | 117.5 | 119.0 | 112.5 | 116.6 | 112.5 | 119.0 | 5.8% |
| 6 | 120.0 | 116.1 | 117.3 | 119.6 | 121.5 | 115.0 | 120.3 | 115.0 | 121.5 | 5.6% |
| 7 | 123.3 | 120.0 | 121.1 | 121.5 | 123.9 | 117.5 | 124.0 | 117.5 | 124.0 | 5.6% |
| 8 | 126.7 | 124.1 | 125.1 | 123.1 | 126.3 | 120.0 | 127.7 | 120.0 | 127.7 | 6.4% |
| 9 | 130.0 | 128.6 | 129.4 | 124.6 | 128.6 | 122.5 | 131.2 | 122.5 | 131.2 | 7.1% |
| 10 | 133.3 | 133.3 | 134.1 | 125.9 | 130.9 | 125.0 | 134.7 | 125.0 | 134.7 | 7.8% |
| 11 | 136.7 | 138.5 | 139.1 | 127.1 | 133.2 | 127.5 | 138.2 | 127.1 | 139.1 | 9.4% |
| 12 | 140.0 | 144.0 | 144.4 | 128.2 | 135.4 | 130.0 | 141.5 | 128.2 | 144.4 | 12.6% |
| 13 | 143.3 | 150.0 | 150.2 | 129.2 | 137.6 | 132.5 | 144.7 | 129.2 | 150.2 | 16.2% |
| 14 | 146.7 | 156.5 | 156.5 | 130.2 | 139.7 | 135.0 | 147.9 | 130.2 | 156.5 | 20.2% |
| 15 | 150.0 | 163.6 | 163.3 | 131.1 | 141.7 | 137.5 | 150.9 | 131.1 | 163.6 | 24.8% |
All values in kg, rounded to 0.1. Because every formula is proportional to the weight lifted, the spread column holds for any load: 60 kg, 225 lb or 140 kg all give the same percentages.
Three zones show up in that last column.
1 to 3 reps: a fixed gap set mostly by Mayhew. The spread runs 7.7% to 8.9%, but most of it comes from one formula. Drop Mayhew and the other six agree within 3.3% at 1 rep and 5.4% at 3 reps (105.9 to 111.6). At these rep counts you are already close to your max, so the estimate matters least.
4 to 10 reps: the tight band. The spread bottoms out at 5.6% for 6 and 7 reps and never goes above 7.8%. On a 100 kg set of 6, the seven answers fall between 115.0 and 121.5 kg. That is 6.5 kg of disagreement, a little more than one pair of 2.5 kg plates.
11 to 15 reps: the fan. The spread goes 9.4%, 12.6%, 16.2%, 20.2% and 24.8%. At 15 reps the formulas offer anything from 131.1 to 163.6 kg for the same performance, a 32.5 kg range. No amount of care in choosing a formula fixes that, because the formulas are disagreeing about something the set never measured.
The same table read backwards: what percentage each formula assigns to a rep count
Coaches often use the inverse question: if a program says "work at 75%", how many reps should that allow? Each formula implies an answer. These are the weight lifted as a percentage of the estimated max:
| Reps | Epley | Brzycki | Lander | Lombardi | Mayhew | O'Conner | Wathan |
|---|---|---|---|---|---|---|---|
| 1 | 96.8% | 100.0% | 98.6% | 100.0% | 91.9% | 97.6% | 98.7% |
| 3 | 90.9% | 94.4% | 93.3% | 89.6% | 87.7% | 93.0% | 91.8% |
| 5 | 85.7% | 88.9% | 87.9% | 85.1% | 84.0% | 88.9% | 85.8% |
| 8 | 78.9% | 80.6% | 79.9% | 81.2% | 79.2% | 83.3% | 78.3% |
| 10 | 75.0% | 75.0% | 74.6% | 79.4% | 76.4% | 80.0% | 74.2% |
| 12 | 71.4% | 69.4% | 69.2% | 78.0% | 73.9% | 76.9% | 70.7% |
| 15 | 66.7% | 61.1% | 61.2% | 76.3% | 70.6% | 72.7% | 66.3% |
So "10 reps at 75%" is an Epley and Brzycki convention. Lombardi says a 10-rep set is 79.4% of your max, O'Conner says 80.0%. At 15 reps the formulas place you anywhere from 61.1% to 76.3% of your max. That 15-point range explains why one lifter's "12 to 15 rep" sets feel easy and another's feel crushing at the same printed percentage: the relationship between load and reps to failure is not the same for everyone, and the formulas encode different guesses about it.
Epley vs Brzycki, rep by rep
These two dominate calculators and gym apps, and they disagree in a predictable way. Epley is a straight line: each extra rep adds one-thirtieth of the weight. Brzycki is a hyperbola: each extra rep adds a little more than the one before, and the formula blows up as reps approach 37, where its denominator reaches zero.
| Reps | Epley (100 kg) | Brzycki (100 kg) | Epley minus Brzycki | As % of Brzycki |
|---|---|---|---|---|
| 1 | 103.3 | 100.0 | +3.3 | +3.3% |
| 3 | 110.0 | 105.9 | +4.1 | +3.9% |
| 5 | 116.7 | 112.5 | +4.2 | +3.7% |
| 6 | 120.0 | 116.1 | +3.9 | +3.3% |
| 8 | 126.7 | 124.1 | +2.5 | +2.0% |
| 10 | 133.3 | 133.3 | 0.0 | 0.0% |
| 12 | 140.0 | 144.0 | −4.0 | −2.8% |
| 13 | 143.3 | 150.0 | −6.7 | −4.4% |
| 15 | 150.0 | 163.6 | −13.6 | −8.3% |
Below 10 reps Epley is always the more generous of the two, by 2% to 3.9% for sets of 3 to 8. At exactly 10 the two give 133.3 kg. Above 10 Brzycki pulls ahead and the gap widens quickly: 2.8% at 12 reps, 8.3% at 15.
For a US lifter, the same comparison in pounds on a 225 lb bench set:
| 225 lb for | Epley | Brzycki | Lowest of 7 | Highest of 7 |
|---|---|---|---|---|
| 3 reps | 247.5 | 238.2 | 238.2 (Brzycki) | 256.5 (Mayhew) |
| 5 reps | 262.5 | 253.1 | 253.1 (Brzycki and O'Conner) | 267.8 (Mayhew) |
| 8 reps | 285.0 | 279.3 | 270.0 (O'Conner) | 287.3 (Wathan) |
| 12 reps | 315.0 | 324.0 | 288.5 (Lombardi) | 324.9 (Lander) |
The 5-rep row matches the worked example on the one rep max calculator: 225 × (1 + 5/30) = 262.5, which the calculator shows rounded to 263 lb.
The calculator also offers an option labelled Wendler. Its code computes weight × reps × 0.0333 + weight, which is Epley written with the rounded constant, so its results sit within a fraction of a pound of the Epley column. It also hides Brzycki above 12 reps and shows Epley instead, which the grid above supports: by 13 reps Brzycki is climbing 6 to 7 kg per rep on a 100 kg set.
Which of the two is "right"? The abstracts below give each one a win in a particular setting, so the answer depends on the lift, the lifter and above all the rep count. For sets of 3 to 8 reps, picking Epley over Brzycki moves your estimate by 2% to 4%. Moving your test set from 8 reps to 13 changes how far the formulas can disagree far more.
Why the curves split after 10 reps
Look again at the shape column in the first table. Below 10 reps all seven curves are close to straight, which is why they agree. Past 10 they bend in different directions.
- The hyperbolas (Brzycki, Lander) bend upward. Each extra rep implies a bigger jump in max strength. Lander's denominator reaches zero just below 38 reps, Brzycki's at 37.
- The straight lines (Epley, O'Conner) keep the same slope forever. Epley adds 3.33% of the load per rep and O'Conner 2.5%, so they drift apart slowly and steadily.
- The exponentials (Mayhew, Wathan) bend downward toward a ceiling. With these constants, Mayhew can never return more than about 1.92 times the lifted weight and Wathan about 2.05 times, however many reps you enter.
- Lombardi's power curve flattens fastest. Going from 10 to 15 reps adds only 5.2 kg on a 100 kg set, against 30.3 kg for Brzycki.
Underneath the math is a physiological point. Reps to failure with a light load is a test of muscular endurance as much as strength, and endurance varies between people with the same max. Reynolds and colleagues tested 70 adults aged 18 to 69 at 1, 5, 10 and 20 repetitions to failure on the bench press and leg press, and the abstract reports that load decreased nonlinearly as the rep number went up. Mayhew and colleagues, training 103 young women for 12 weeks, found the change in reps to fatigue at a fixed percentage varied widely between individuals (27% ± 99%), even though the group average barely moved. A formula is a single average curve; the further you go from a heavy set, the further individual curves wander from it.
What the validation studies found
The studies below each compared estimated with measured one rep maxes. They differ in exercise, population and which formulas they tested, so they cannot be pooled into a single ranking. Each line states only what that study's PubMed abstract reports.
| Study | Who and what | Finding relevant to formula choice |
|---|---|---|
| Mayhew et al., 1995 | 220 men in seven groups, from untrained students to football players; bench press | Six relative-endurance equations gave validity coefficients of 0.82 to 0.98. In men completing 10 or fewer reps, three equations overpredicted and two underpredicted; Brzycki was the most accurate. Above 10 reps, three overpredicted and three underpredicted. Averaging two equations may reduce error. |
| Whisenant et al., 2003 | 69 college football players; 225 lb bench press reps | Across 11 equations, validity was higher when fewer repetitions were completed. |
| Reynolds et al., 2006 | 70 adults aged 18 to 69; bench press and leg press | 5RM data gave the best prediction (R² of 0.993 for chest press and 0.974 for leg press); no more than 10 repetitions should be used in linear equations to estimate 1RM. |
| Kemmler et al., 2006 | 70 trained postmenopausal women aged about 57; four machine exercises | A cubic equation predicted 1RM from 3 to 20 reps with mean absolute differences of 1.5% to 3.1%, and accuracy did not depend on exercise or rep count. |
| Mayhew et al., 2007 | 118 high-school male athletes; bench press | 7- to 10-RM equations correlated above 0.98 with actual 1RM, but only 3 of 11 equations gave values not significantly different from actual. The best put 65% of athletes within ±4.5 kg. |
| Mayhew et al., 2008 | 103 young women before and after 12 weeks of training; bench press | Prediction equations were more accurate when fewer than 10 reps to fatigue were used, both before and after training. |
| McNair et al., 2011 | 18 people with knee osteoarthritis; knee extension and leg press machines | 12 equations compared. Knee extension: Brown, Brzycki, Epley, Lander, Mayhew et al., Poliquin and Wathen were most accurate (typical error 3% to 4%). Leg press: Adams, Berger, Kemmler et al. and O'Conner et al. were most accurate (5.9% to 6.3%). |
| Ribeiro Neto et al., 2017 | 45 men with spinal cord injury; bench press | No significant differences between measured 1RM and the existing equations. Lombardi had the best Bland-Altman results (0.5 kg mean difference). |
| García-Ramos et al., 2019 | 26 men (22 rowers, 4 weightlifters); prone bench pull | The Lombardi and O'Connor equations overestimated 1RM by 3.43% to 4.08%. |
| Nickerson et al., 2020 | 20 resistance-trained adults; bench press reps to failure at 70% of 1RM | Epley gave smaller mean differences than Wathen and Mayhew and the lowest standard error of the methods tested; the authors recommend Epley over Wathen, Mayhew and a velocity-based method when loads up to 70% of 1RM are used. |
| Pérez-Castilla et al., 2021 | 23 healthy men and women; lat pull-down and seated cable row at 80% of 1RM | The Mayhew and Wathen equations significantly underestimated 1RM (by 2.14 to 6.65 kg); velocity-based methods were more accurate. |
Three things stand out.
Fewer reps keeps coming up. Whisenant 2003, Reynolds 2006 and Mayhew 2008 tie accuracy to a lower rep count. Mayhew 1995 split its men at 10 reps and found every equation significantly biased above that line, with Brzycki the most accurate at 10 or fewer. That is the same threshold where the grid above starts to fan out. Different evidence, same conclusion.
There is no overall winner. Brzycki comes out best in one bench press study and Epley in another. Lombardi does best in men with spinal cord injury, yet overestimates on the prone bench pull. On knee extension, the accurate group includes Brzycki, Epley, Lander, Mayhew and Wathen together; on leg press in the same people, a different set comes out ahead. Anyone telling you one formula is "the most accurate" is generalising from one of these settings.
Errors run in both directions. The same equations overestimated on a prone bench pull and underestimated on lat pull-downs and cable rows. You cannot fix a formula's bias for your lift by adding or subtracting a fixed amount without measuring it yourself.
LeSuer and colleagues' 1997 comparison of the seven formulas on the bench press, squat and deadlift is the study most often quoted for these equations, and it is the source of the formula table used here. It has no abstract on PubMed, so its accuracy findings are not summarised on this page; every finding in the table above can be checked against its PubMed abstract.
Choosing a formula for your lift
Since no formula wins everywhere, a sensible choice depends on what the studies tested and on what you need the number for.
| Your situation | What the evidence above supports |
|---|---|
| Barbell bench press, trained or untrained men, set of 10 or fewer | Brzycki was the most accurate of six in Mayhew 1995 |
| Bench press set at about 70% of your max (12 to 13 reps by the Epley curve) | Epley outperformed Wathen and Mayhew in Nickerson 2020; this is the one study here that supports a set past 10 reps, and only for Epley on bench press |
| Machine knee extension | Several formulas performed similarly in McNair 2011, including Brzycki, Epley, Lander, Mayhew et al. and Wathen |
| Machine leg press | O'Conner et al. was among the most accurate in McNair 2011 |
| Pulling exercises (rows, pull-downs) | Rep equations were biased in both directions in the two pulling studies; treat any estimate as rough |
| You want to track progress over months | Any one formula, used consistently with the same rep range, since the bias stays roughly constant for you |
| You can't decide | Average two formulas, as Mayhew 1995 suggests may reduce error. Epley and Brzycki together bracket the middle of the 4 to 10 rep band |
The last row has a property worth noticing. Below 10 reps Epley runs high and Brzycki runs low, and both sit inside the middle of the seven-formula range. Their average on a 100 kg set of 6 is 118.1 kg, against a seven-formula median of 119.6 and a range of 115.0 to 121.5.
For tracking, consistency beats the choice of formula. If you test bench press at 5 reps with Brzycki in January and at 12 reps with Epley in April, the change in your estimate mixes real strength gains with a switch in method. Keep the formula and the rep range fixed, and the trend line means something even if the absolute number carries a few percent of bias.
Which rep range to test: a rule
Combining the grid with the validation studies gives a rule you can apply at the gym:
Pick a load you expect to fail with between 3 and 10 clean reps, aim for about 5, take it to technical failure, and only run the numbers if you landed in that window.
Why each part is there:
- 3 to 10 reps. Between 4 and 10 reps the seven formulas agree within 7.8%, and the studies that looked at rep count favour 10 or fewer. Three reps is included because the estimate is close to your true max anyway; below that, a formula is barely needed.
- About 5. Reynolds 2006 found 5RM gave the best prediction of the counts it tested (5, 10 and 20). In the grid, 5 to 7 reps is also where the formulas cluster tightest.
- Technical failure. These are repetitions-to-failure equations. Stopping at 8 reps when you had 10 in you will underestimate your max by whatever the formula assigns to those 2 missing reps: about 6.7 kg on a 100 kg set with Epley. Count the last rep you completed with full range and the form you'd use in training.
- Only run the numbers in the window. If you got 13, the honest output is a range, not a number: 129 to 150 kg on a 100 kg set. Add load and re-test another day rather than picking the formula that gives the answer you hoped for.
What to do with each outcome:
| Reps you got | What the result is worth | Next step |
|---|---|---|
| 1 to 2 | Close to your max already; formulas differ mainly by their 1-rep offset | Use the weight itself, or Brzycki or Lombardi, which return it unchanged at 1 rep |
| 3 to 10 | Formulas agree within about 6% to 8%; this is the usable window | Use one formula consistently, or average Epley and Brzycki |
| 11 to 12 | Spread 9% to 13%; usable as a rough guide | Note it as an estimate, re-test heavier when convenient |
| 13 to 15 | Spread 16% to 25%; formulas disagree by more than a typical training increment | Do not program off it; re-test with a heavier load |
| More than 15 | Outside the range most of these formulas were tested over | Treat it as an endurance result, not a strength test |
A note on setting: the reps you get on a fresh first set differ from the reps you get as a third set after squats. Test when you are fresh, after a normal warm-up, on the same equipment you will train with. A machine, a Smith bar and a free barbell will not give interchangeable numbers, and McNair 2011 shows that even the best formula can change between two machines used by the same people.
A worked example from one set
Take a hypothetical lifter who squats 80 kg for 6 clean reps to failure. The seven formulas give:
| Formula | Estimate from 80 kg × 6 |
|---|---|
| O'Conner | 92.0 kg |
| Brzycki | 92.9 kg |
| Lander | 93.8 kg |
| Lombardi | 95.7 kg |
| Epley | 96.0 kg |
| Wathan | 96.3 kg |
| Mayhew | 97.2 kg |
The range is 92.0 to 97.2 kg, a gap of 5.2 kg. Averaging Epley and Brzycki gives (96.0 + 92.9) / 2 = 94.45 kg, which we'll call 94.5 kg. Training loads at common percentages of that figure:
| Target | Calculation | Load | Nearest 2.5 kg increment |
|---|---|---|---|
| 70% | 94.45 × 0.70 | 66.1 kg | 65 kg |
| 75% | 94.45 × 0.75 | 70.8 kg | 70 kg |
| 80% | 94.45 × 0.80 | 75.6 kg | 75 kg |
| 85% | 94.45 × 0.85 | 80.3 kg | 80 kg |
Notice the 85% row lands on the test weight itself, which tells you something useful: by these formulas, a set of 6 to failure sits near 85% of the max. That matches the percentage table above (Epley says 6 reps is 83.3%, Brzycki 86.1%), and it gives the lifter a quick sanity check. If 80 kg felt like a hard 6, programming several sets of 5 at 80 kg is close to maximal work, not a moderate session.
Now suppose the same lifter had instead used 65 kg and ground out 14 reps. The formulas would place the max anywhere from 84.6 kg (Lombardi) to 101.7 kg (Brzycki and Lander), a 17.1 kg range, against the 92.0 to 97.2 kg range from the heavier set. One light set has turned a 5 kg question into a 17 kg one, and the answer now depends mostly on which formula you happened to open. That is the practical cost of testing too light, and it is why the rule above stops at 10.
The percentage calculator handles the training-load arithmetic if you would rather not multiply by hand.
When an estimate is the better test
A true one rep max test means warming up through several heavy singles, needing a spotter or safety bars, and accepting a very heavy, high-effort lift. For many lifters that is a fine thing to do occasionally. For others, the reps-to-failure estimate is the better tool, and some of the study populations above show why researchers used it.
Abdul-Hameed and colleagues tested estimated 1RM (with the Brzycki equation) in 25 untrained middle-aged adults with type 2 diabetes. Their abstract notes that the safety of actual 1RM testing is questionable in clinical situations such as type 2 diabetes, and reports excellent test-retest reliability for the estimate (ICC of 0.98 to 0.99) across five exercises. McNair 2011 (knee osteoarthritis), Ribeiro Neto 2017 (spinal cord injury) and Kemmler 2006 (trained postmenopausal women) all found prediction equations usable in their groups.
That is not medical advice, and none of those studies says who should or should not lift heavy. If you have a heart, joint, blood-pressure or other health condition, or you are new to resistance training, ask a doctor or a qualified exercise professional before testing near your limit, whether by a single or by reps to failure. A reps-to-failure set still ends in a maximal effort; it just happens at a lighter load.
Two other alternatives turned up in the abstracts. Velocity-based estimates, which track how fast the bar moves at several submaximal loads, beat the Mayhew and Wathen equations on two pulling exercises in Pérez-Castilla 2021, though in Nickerson 2020 the velocity method had the largest standard error on bench press. Both need a linear position transducer or a phone app and some care in setup. For most gym users, a well-chosen rep test and a consistent formula remain the practical choice.
Re-testing after a training block
An estimate is a snapshot. Mayhew 2008 found that, after 12 weeks of training, women's bench press max rose 28% on average while their reps to fatigue at the same relative load changed very differently from person to person. In plain terms: your load-to-reps curve can shift as you train, so a formula that fitted you well in month one may not in month four.
A practical rhythm:
- Test at the start of a training block using the rep-range rule above, and note the load, the reps and the formula.
- Train from the estimate for 4 to 8 weeks, adjusting loads by feel when sets are clearly too easy or too hard.
- Re-test with the same exercise, equipment, warm-up and formula, aiming again for 3 to 10 reps, ideally near 5. Because your max has likely gone up, use a heavier test load than last time so you don't drift into the 12 to 15 rep zone.
- Compare the two estimates. For a lifter testing around 80 kg, one extra rep moves the Epley estimate by about 2.7 kg, so a change smaller than that is within the effect of a single rep and is best read as no change.
If you also track bodyweight or body composition, an estimate per kilogram of lean mass can be more informative than the raw number; the lean body mass calculator gives that figure, and the protein calculator sets intake targets for a training block. Energy needs for heavier training are covered in the BMR equations comparison, which takes the same approach to a different set of competing formulas. More tools for training and conditioning are collected in the fitness hub.
When you have your next set of numbers, enter them in the one rep max calculator with the formula you settled on here; switching between its four options on the same entry shows the spread for your own set in a few seconds.
Sources
Formula definitions read from the Scientific Reports table (reproducing LeSuer et al., 1997) and the reference list of a PeerJ paper, both on PubMed Central. Study findings read from PubMed abstracts. All checked October 30, 2026; grids and worked examples computed by CalculatorVillage on October 30, 2026.
- Scientific Reports, 2021: The effects of barbell resistance exercise on information processing speed and conflict-related ERP in older adults, Table 1 (full text on PubMed Central): the seven prediction equations as given by LeSuer et al. (1997).
- PeerJ, 2023: The reliability and validity of repeat power ability assessments and measurement indices in loaded vertical jumps (full text on PubMed Central): reference list giving the source years for Epley, Lander, Lombardi, O'Connor, Mayhew, Brzycki and Wathen.
- Mayhew JL et al., 1995: Muscular endurance repetitions to predict bench press strength in men of different training levels.
- Whisenant MJ et al., 2003: Validation of submaximal prediction equations for the 1 repetition maximum bench press test on a group of collegiate football players.
- Reynolds JM et al., 2006: Prediction of one repetition maximum strength from multiple repetition maximum testing and anthropometry.
- Kemmler WK et al., 2006: Predicting maximal strength in trained postmenopausal woman.
- Mayhew JL et al., 2007: Using absolute and relative muscle endurance to estimate maximal strength in young athletes.
- Mayhew JL et al., 2008: Accuracy of prediction equations for determining one repetition maximum bench press in women before and after resistance training.
- McNair PJ et al., 2011: Predicting maximal strength of quadriceps from submaximal performance in individuals with knee joint osteoarthritis.
- Abdul-Hameed U et al., 2012: Reliability of 1-repetition maximum estimation for upper and lower body muscular strength measurement in untrained middle aged type 2 diabetic patients.
- Ribeiro Neto F et al., 2017: Validity of one-repetition maximum predictive equations in men with spinal cord injury.
- García-Ramos A et al., 2019: Reliability and validity of different methods of estimating the one-repetition maximum during the free-weight prone bench pull exercise.
- Nickerson BS et al., 2020: Evaluation of load-velocity relationships and repetitions-to-failure equations in the presence of male and female spotters.
- Pérez-Castilla A et al., 2021: Validity of different velocity-based methods and repetitions-to-failure equations for predicting the 1 repetition maximum during 2 upper-body pulling exercises.