How Accurate Are 1RM Formulas? Epley vs Brzycki

You lifted 225 lb for 5 reps and your app told you your one-rep max is 253 lb. You never lifted 253 lb, so how much should you trust that number? Rep-max formulas like Epley and Brzycki turn a set you did do into an estimate of a single max you didn’t. They are useful, they are fast, and they are genuinely good within a narrow window, but they are estimates, not measurements. This is a look at how the two most common formulas work, where they agree, and where the error creeps in. If you just want the number, the one-rep max calculator runs several formulas at once so you can see the spread.

The two formulas, side by side

Both formulas take the same two inputs: the weight you lifted and the number of reps you got with it, ideally to or near failure.

Epley: 1RM = weight × (1 + reps ÷ 30)

Brzycki: 1RM = weight × 36 ÷ (37 − reps)

That’s the whole thing. Plug in 225 lb for 5 reps:

• Epley: 225 × (1 + 5/30) = 225 × 1.1667 = 262.5 lb
• Brzycki: 225 × 36 ÷ (37 − 5) = 225 × 1.125 = 253.1 lb

A 9 lb gap on the same set. Neither is “wrong”; they’re built from different curves fit to different data. The math here is deterministic, so you can always reproduce it by hand.

Where the two formulas cross

Epley and Brzycki are not rivals across the whole range. They intersect at exactly 10 reps, where both return weight × 1.3333. Below 10 reps, Epley predicts a slightly higher max; above 10 reps, Brzycki pulls ahead. The table shows the estimated 1RM each formula gives for a fixed 100 lb working set at different rep counts.

Reps to failure	Epley estimate	Brzycki estimate	Gap
1	103 lb	100 lb	3 lb
3	110 lb	106 lb	4 lb
5	117 lb	113 lb	4 lb
8	127 lb	124 lb	3 lb
10	133 lb	133 lb	0 lb
12	140 lb	144 lb	4 lb
15	150 lb	164 lb	14 lb

The pattern is the real lesson: at low rep counts the two formulas are within a few percent of each other, and they converge on 10. Out past 12–15 reps they diverge sharply, and both become unreliable. That’s not a quirk of these two equations; it’s the central finding of the research on rep-max prediction.

What the research shows

Study after study lands on the same three points.

Correlation is high, but that hides real error. Predicted and actual maxes correlate tightly, usually r > 0.90, sometimes above 0.95. That sounds like near-perfection, but a high correlation only means the estimates rank people correctly; it doesn’t mean any single estimate is dead-on. Absolute errors in the range of roughly 3–7% are common even in the good conditions, which on a 300 lb lift is 9–21 lb.

Fewer reps means less error. Accuracy is best when the set used is short. A 2024 cross-validation analysis of bench press and leg extension found the most reliable predictions came from sets ending at roughly 4–10 reps to failure, and accuracy degraded meaningfully beyond about 10 reps. The intuition is simple: a set of 3 is physiologically close to a true max, so there’s little curve to extrapolate. A set of 20 is a different kind of effort, muscular endurance, and forcing it onto a strength curve introduces large error.

The lift matters. The formulas were mostly built on bench press data, and they travel imperfectly. Compound lower-body lifts, above all the deadlift, tend to be predicted less accurately, and several datasets show equations systematically underestimating the deadlift. Velocity-based work in the squat and bench presses reaches a similar conclusion: no single equation is universally best across every movement.

The honest summary: within about 10 reps, a rep-max formula will usually land within a few percent of your true max. Push the reps higher and the estimate becomes a guess dressed up as a number.

Epley or Brzycki: which should you use?

For most people, in the rep ranges you should be testing, it barely matters: under about 8 reps the two agree within a couple of percent. If you want a rule of thumb: Brzycki is often favored for low-rep sets (1–6), Epley for moderate sets (6–10), and neither should be trusted much past 12. The single most useful habit is to pick one formula and stick with it. Because you’re tracking change over time, a small consistent bias cancels out. A 253 lb estimate that becomes a 268 lb estimate is a real gain regardless of which formula drew the curve. Switching formulas between sessions just adds noise. Building strength this way is one of the most durable things you can do for healthspan; see our overview of strength training for longevity for why the trend line matters more than any single test.

How to get the most accurate estimate

• Keep the set short. Use a weight that brings you to failure, or one rep shy of it, somewhere in the 3–8 rep range. This is the sweet spot where the formulas were validated best.
• Approach failure for real. The math assumes the rep count reflects genuine near-maximal effort. Stopping with four reps in the tank makes the input meaningless.
• Use a compound, stable lift. Bench and squat estimate more cleanly than the deadlift or single-joint movements.
• Warm up and control the tempo. Grinding, cheated reps inflate the count and the estimate.
• Treat it as a range, not a verdict. Running the same set through several formulas gives you a low-to-high band, which is more honest than a single decimal-precise figure.

The takeaway

Epley and Brzycki are two slightly different curves fit to the same real-world question, and for practical training they agree far more than they disagree, above all where you should be testing, at 10 reps or fewer. Understanding the limits of 1RM formula accuracy keeps you from over-reading a number the formula was never precise enough to promise. Estimate from a short, honest set, pick one formula, and watch the trend. Run your numbers through the one-rep max calculator to see how the formulas compare on your own lifts.

Sources

• Cross-Validation of Equations for Estimating 1RM From Repetitions to Failure for the Bench Press and Leg Extension. J Strength Cond Res, 2024 (PMID 39495260)
• Accuracy of Predicting One-Repetition Maximum from Submaximal Velocity in the Barbell Back Squat and Bench Press. 2022 (PMC9465738)
• Developing Accurate Repetition Prediction Equations for Trained Older Adults with Osteopenia. 2024 (PMC11435939)
• LeSuer DA, McCormick JH, Mayhew JL, et al. The accuracy of prediction equations for estimating 1-RM performance in the bench press, squat, and deadlift. Journal of Strength and Conditioning Research, 1997;11(4):211–213.