How the 1RM estimate works
The calculator runs seven formulas side by side: Epley, Brzycki, Lombardi, O'Conner, Mayhew, Wathan, and Lander. Each formula was built from a different data set and a different mathematical fit, which is why the answers can differ by a few pounds for the same set. e1RM shows the spread instead of hiding it behind one default number.
The spread shows how much the models disagree for the weight and rep count you enter. It is not an error bar or a measured confidence interval for your true max. A narrow spread does not prove accuracy, and a wide spread cannot tell whether you were fatigued, far from failure, or using inconsistent technique. Those details require information about the actual set.
The seven formulas
Epley (1985) uses 1RM = w x (1 + r / 30). It is the most common default in strength training apps and works well for hard sets of roughly two to eight reps. It can read high when a lifter is unusually good at rep work.
Brzycki (1993) uses 1RM = w × 36 / (37 − r). It is usually a little more conservative than Epley. That makes it useful when returning from a layoff, choosing a training max for a high-volume block, or estimating from a set that had a rep in reserve.
Lombardi (1989) uses 1RM = w x r^0.10. It is a power-curve fit rather than a simple linear adjustment. It often stays close to the middle of the group for low-rep barbell work.
O'Conner (1989) uses 1RM = w x (1 + 0.025 x r). It is a simple linear approximation and often lands near Brzycki for common working sets.
Mayhew (1992) uses 1RM = (100 x w) / (52.2 + 41.9 x e^(-0.055 x r)). It came from bench press research with athletic populations, so it is useful as an upper-body cross-check.
Wathan (1994) uses 1RM = (100 x w) / (48.8 + 53.8 x e^(-0.075 x r)). It is another curved model designed to handle common strength-test rep ranges without assuming every extra rep adds the same amount.
Lander (1985) uses 1RM = (100 x w) / (101.3 - 2.67123 x r). It is a denominator-based model that often lands between conservative formulas such as Brzycki and more liberal formulas such as Epley.
In every formula, w means weight lifted and r means reps completed. The calculator keeps the math visible because lifters should be able to cite the formula, check the number, and understand why two calculators may disagree.
Which formula should I trust?
No formula is guaranteed to match every lifter or exercise. Compare the estimates, then check them against your own performance using a consistent movement standard. Epley and Brzycki offer different mathematical estimates; choosing the lower number does not make it a validated lower bound for your true max.
For example, 225 x 5 gives 262.5 lb with Epley and approximately 253.1 lb with the Brzycki equation used here. The difference comes from the equations. It does not show how far either result is from your actual max, and your actual max can fall outside the displayed range. Use the result as a planning estimate rather than a guaranteed lift.
Why low-rep sets give better estimates
A clean triple at roughly 90% of true max is mechanically similar to a heavy single. Bar path, bracing, joint angles, and intent are close to max-strength work. A 12-rep set at 65% is different: breathing, pacing, local muscular endurance, and pain tolerance all become limiting factors. That is why high-rep estimates often inflate a lifter's usable max.
The sweet spot is usually two to six reps. Sets of seven to ten can still work, but the error bar widens. Anything above ten reps should be treated as a rough estimate until you confirm it with a heavier set.
From estimated 1RM to training loads
The percentage table converts the selected formula into practical barbell loads. Use 95-100% for singles and max-strength exposure, 85-92.5% for heavy strength work, 75-82.5% for strength and hypertrophy overlap, 65-72.5% for volume and technique, and 50-60% for speed work, warm-ups, deloads, or AMRAP-style variations. For a repeatable block, first turn the estimate into an 85-95% training max, then review the full 1RM percentage chart.
A common mistake is plugging an inflated estimate into 85% work and grinding through bad reps for three weeks. If your first set at 85% moves like an RPE 9.5 max effort, the estimate was too high. Drop the training max by 5%, retest with a cleaner set, and rebuild the percentage table.
When to retest
Intermediates can retest every four to eight weeks without taking a true max. Use a top set of three or five reps, keep the standard consistent, and compare the new estimate to the old one. A cleaner set with the same weight is progress even when the formula output barely moves.
Estimated maxes are best used as planning tools. They are not meet attempts, they are not proof of a gym PR, and they are not a substitute for judgment. Comparing seven formulas makes model differences visible; it does not measure the quality of your set or your individual prediction error. Once the load is chosen, use the plate calculator and warm-up set calculator to make the number usable in the gym.
How to choose the input set
Use a set that matches the way you want to train the lift. For bench press, decide whether the reps were paused or touch-and-go. For squat, keep depth honest. For deadlift, note whether straps were used. For overhead press, keep strict press and push press separate. The calculator can only estimate from the information you give it, so a consistent movement standard matters more than chasing the hardest-looking set in your logbook.
Do not enter warm-up sets, backoff sets taken far from failure, or sets where technique changed dramatically across reps. A clean 225 x 5 at RPE 9 is a better input than 225 x 8 with three questionable reps. If you are unsure, calculate both the aggressive and conservative set, then program from the lower training max until performance confirms the higher one.
Sources
- Brzycki 1993 - Strength testing: predicting a one-rep max from reps-to-fatigue.
- Mayhew et al. 1992 - Relative muscular endurance performance as a predictor of bench press strength.
- LeSuer et al. 1997 - The accuracy of prediction equations for estimating 1-RM performance.
