RPE and RIR explained: the reps-in-reserve scale, how accurate it is, and how to set load with it

What RPE and RIR mean in lifting

RIR is the number of reps you could still have done when you stopped a set; RPE is the same judgement on a 1-to-10 scale, where RPE 10 means none were left. Stop a set of squats after eight reps knowing you had two more in you, and that set was 2 RIR, or RPE 8. Both describe how close the set came to failure, not how heavy the bar was.

That is the difference from a percentage. "80% of your one-rep maximum" fixes the load and lets the effort vary from day to day. "5 reps at RPE 8" fixes the effort and lets the load vary. The second is called autoregulation: on a good day the bar goes up, on a tired day it comes down, and the set still ends about two reps short of failure.

A set has three numbers, and RIR links them: reps done + reps in reserve = reps possible at that load. Helms and colleagues give the example that six reps with 0 RIR, five with 1 RIR, four with 2 RIR and three with 3 RIR are roughly the same effort, because each describes a load you could lift about six times (Strength and Conditioning Journal, 2016).

The RPE to RIR table

On the lifting scale, RPE 10 is 0 reps in reserve, 9 is 1, 8 is 2 and 7 is 3; below 7 the descriptions turn into ranges and then into words. The whole-number pairs are from Zourdos and colleagues' paper that introduced the scale (J Strength Cond Res 30(1):267-275, 2016). The half-points are quoted from the methods of Helms and colleagues' 2018 training study (Frontiers in Physiology 9:247), and the low end from the 2016 practical article by the same group.

The scale uses reps in reserve only from 5 upward. Helms and colleagues explain that below that, a count of reps left is too far from failure to estimate, so 1 to 4 are rated by effort words instead, and scores of 5 and 6 were grouped as 4 to 6 reps left.

Where the scale came from: Borg's scales and the 2016 lifting scale

The lifting scale borrowed the name RPE from Gunnar Borg's exertion scales but replaced how-hard words with a count of reps left. Borg's original scale, built for whole-body work such as cycling, rates exertion from 6 to 20, a range chosen to roughly match heart rate; his CR10 scale followed, rated on numbers up to 10 (Borg, Medicine and Science in Sports and Exercise, 1982). Helms and colleagues' 2016 review traces that line, the 6-to-20 range and the later picture-based OMNI scale.

The reps-in-reserve version came from powerlifting coaching. The same review says a scale of this type "was originally created in 'The Reactive Training Systems Manual' in 2008", a manual by Tuchscherer (as the review cites it), before Zourdos and colleagues tested it. Their 2016 study had 29 people squat a one-rep maximum, single reps at 60, 75 and 90% of it, and a set of eight at 70%, rating each with the new scale. Experienced squatters rated their maximum at 9.80 on average and novices at 8.96, so the novices under-rated a true maximal lift by about one point.

The same number on different scales means different things, and that is the most common mix-up.

The CDC's scale, for walking, running and other activity, puts sitting at 0, moderate effort at 5 or 6 and vigorous effort from 7 or 8. A programme that says "RPE 7" next to a squat means the third row, and one that says it next to a run almost certainly means one of the first two.

How to read "3×5 @ RPE 8"

"3×5 @ RPE 8" means three sets of five reps at a load that leaves about two reps in reserve on each set. The sets-first, reps-second convention is the usual one, explained in the guide to how sets and reps are written; the "@" adds the effort, and the load is left to you.

The "3×10 at RPE 8-9 (i.e., 2 or 1 RIR)" and "5RM with 1 RIR" forms are both examples in Helms and colleagues' 2016 article, which also suggests pairing a percentage with an RIR target, such as "0-1 RIR on all sets", so a lifter adjusts a load that turns out too easy or too hard.

How accurate are lifters at guessing reps in reserve?

Lifters usually guess reps in reserve within one to two reps near the end of a set, and they tend to guess too few: they usually had about one more rep than they said. That is the pooled result of a 2022 meta-analysis by Halperin and colleagues of 13 publications covering 12 studies and 414 people: an underprediction of 0.95 reps (95% confidence interval 0.17 to 1.73), with very large differences between studies (Sports Medicine 52:377-390).

Distance from failure matters more than experience. Across the studies below, guesses improved near failure, in later sets and in shorter sets, while training experience made little or no difference.

The squat study shows the trap in a long set. At 70% of a maximum, the men averaged 16 reps to failure, and when they called RPE 5 early in the set they were wrong by more than five reps on average. The same men were within about two reps by the time they called RPE 9. A set of 15 reps or more at a moderate load is the hardest kind to rate.

Hitting a prescribed rating is a related skill. In a study of 12 powerlifters by Helms and colleagues, the rating after a set differed from the rating the programme asked for by 0.33 RPE on average (J Strength Cond Res 31(10):2938-2943, 2017). Those were experienced competitive lifters, so treat it as what is possible with practice.

Setting the load across sets: worked examples

The simplest published rule is to change the next set's load by 2% for every half-point the set landed outside its target range. Helms and colleagues used it in an eight-week study of trained men; the RPE group's strength gains were similar to a group lifting fixed percentages, with a possible small advantage for RPE loading (Frontiers in Physiology, 2018).

A missed rep counted as RPE 10, and each missed rep as one more full point. The study's own example: a lifter who completed 4 of 5 reps against an RPE 7 to 9 target dropped the next set by 8%, 4% for landing a point above 9 and 4% for the missed rep.

Example 1: a set that is too easy, then one in range

The programme says 3×5 @ RPE 7-9 for the squat, the scheme the study used in weeks 4 and 5.

1. Set 1: 100 kg × 5, rated RPE 6. That is one point below 7, two half-points, so add 4%: 104 kg.
2. Set 2: 104 kg × 5, rated RPE 8. In range, so keep the load.
3. Set 3: 104 kg × 5, rated RPE 9. Still in range. The session's working load was 104 kg.

Example 2: a missed rep

Same programme, a tired day. Set 1 at 104 kg reaches four reps and the fifth fails. The set counts as RPE 10, one point above 9 (4% off), plus one missed rep (another 4% off): 8% lighter. 104 × 0.92 = 95.7 kg, so load the nearest weight you can make below it, 95 kg, for set 2.

Example 3: using reps plus reps in reserve to choose next week's load

This week: 80 kg × 5 @ RPE 9, one rep in reserve, so 80 kg is about a six-rep maximum (5 + 1). Next week's programme says 3×4 @ RPE 8. Four reps plus two in reserve is also six, so by the equivalence in Helms and colleagues' 2016 article, 80 kg is the starting guess. The first set's rating then confirms it or moves it by the 2% rule.

Rounding matters at light loads. 2% of a 20 kg dumbbell is 0.4 kg, smaller than most dumbbell jumps, so on light lifts the practical move is the next available weight, or one more or one fewer rep.

Counting the reps so the rating means something

A rating is attached to a rep count, and a wrong count makes the rating wrong too. "RPE 8" after what you think was five reps but was actually six describes a different load. The count is also how you check yourself: Helms and colleagues suggest rating a submaximal set, then taking a later set at the same load to failure to see whether the rating was right.

The rep counter does the counting part. Set the target, five for a 3×5, as the maximum so the counter stops there, tap or press Space once per rep, and keep a second counter for sets. The rating is yours to judge at the end of the set; write it next to the number the counter shows. The rep schemes table has the counter settings for 5×5, 3×8-12 and other common schemes.

For a test set to failure, turn the limits off so the counter keeps going past the target. The number it shows minus the reps you did in the rated set is the true reps in reserve at that load. If you guessed 2 and the test shows 3, you are underpredicting by one rep, which is the average error the meta-analysis found.

Ratings also feed volume. A hard set, in the training volume calculator, is one finished with 0 to 3 reps left, about RPE 7 or more, so a correct rating decides which sets count toward the weekly total.

How close to failure: what the studies found

For muscle growth, sets ending at or a few reps short of failure all work, and the evidence on exactly how close is mixed; for strength, the distance from failure matters less. A 2023 meta-analysis of 15 studies by Refalo and colleagues found a small overall effect favouring failure over non-failure (0.19, confidence interval 0.00 to 0.37) that disappeared when failure was defined strictly as momentary failure (Sports Medicine 53:649-665).

A 2024 series of meta-regressions by Robinson and colleagues estimated how many reps short of failure each study's sets ended, and found muscle growth rose as sets ended closer to failure, while strength gains showed no clear relationship (Sports Medicine 54:2209-2231). The authors call the analysis exploratory, because the reps in reserve were estimated from how each study described its training.

For muscle growth, Helms and colleagues' 2016 article advises mostly 6 to 12 reps at RPE 8 to 10 depending on the phase of training, main lifts such as squats and bench press mostly at RPE 6 to 8, and sets to failure, RIR 0, limited to the last set for a body part, mainly on simple isolation exercises. How many sets and reps to do is a separate question, covered in the guide to rep ranges and rest.

What this guide does not cover

This guide explains a rating scale and what published studies found about it; it is not medical or training advice for any individual. Most of the studies are of healthy young adults, many of them experienced lifters, and most tested barbell squat and bench press or machines. If you are new to lifting, have an injury, a heart or blood-pressure condition, or are pregnant, ask a doctor or a qualified coach before taking sets close to failure.

Beginners can still use the scale as a habit. The authors who introduced it advise novices to practise recording reps in reserve "but likely not base training intensity or progression solely" on it. Simple Online Counter is a free, browser-based set of counters; its rep counter counts reps and sets in the browser and keeps them on your device, and the rating at the end of each set is yours.