RPE Chart: RPE to Reps in Reserve and Load

RIR is roughly 10 minus RPE, and reps to failure is reps completed plus that difference — which is all you need to turn a rated set into an estimated 1RM and a load for any other rep target. The chart is most trustworthy near failure and at low rep counts.

Published Last updated 8 sources

Key takeaways

  • The lifting RPE chart is one rule plus a table: RIR ≈ 10 − RPE [1].
  • Reps to failure = reps completed + (10 − RPE). That single line is what turns an RPE into a load estimate.
  • Load estimates from a set are models, not measurements — they degrade as the rep count rises [6].
  • Trust the chart most near failure, where your own estimate of reps left is most accurate [3].

This is the reference page. If you want the reasoning behind the scale, the accuracy data and where RPE breaks down, start with the full RPE guide.

RPE to reps in reserve

The resistance-training RPE scale is anchored to reps in reserve, with RPE 10 meaning no reps left [1]. It exists to let you prescribe and adjust load by effort rather than by a fixed percentage [2].

The two columns are interchangeable by definition; RPE vs RIR covers which of them is the better thing to write into a program.

RPE to reps-in-reserve conversion table
RPEReps in reserveFeels likeWhat it meansTypical use
100Max effortNo reps left. A true limit set.Testing, rare top sets
9.50.5Almost maxCould maybe add load, definitely not another rep.Peaking singles
91Very hardAbout one rep left. Another rep would be questionable.Heavy top sets
8.51.5Hard+Between one and two reps left.Strength work
82HardAbout two reps left. Strong effort, not a grind.Most working sets
73ToughAbout three reps left. Demanding but controlled.Volume work, technique
64ChallengingAbout four reps left. Good working effort.Accessories, deloads
55SteadyAbout five reps left.Deloads, high-fatigue days
46ModerateNoticeable effort, nowhere near a limit set.Warm-up ramp
37LightControlled work, plenty of reps available.Warm-up ramp
28EasyBarely challenging.Warm-up
19Very easyWarm-up effort. Many reps in reserve.Warm-up

Rule of thumb: RIR ≈ 10 − RPE. Half-point steps exist mainly near failure, where lifters can still tell "one more rep" from "maybe one more rep".

Filter to working sets or whole numbers; tap a row to hold it. The wording matches the scale shipped in the Reppit app [S10].

Turn an RPE into a load

Any set gives you two numbers that matter: the reps you did and how many you had left. Add them and you have reps to failure, which is what every 1RM formula actually wants.

Estimated 1RM: 121.7 kg (from 100 kg × 5 reps @ RPE 8 7 reps to failure)

Suggested loads (kg) for each rep target and RPE
RepsRPE 6RPE 7RPE 8RPE 9RPE 10
1106.2109.4112.7116.1121.7
3100103.1106.2109.4112.7
594.197100103.1106.2
885.788.491.394.197
1080.38385.788.491.3
1274.977.680.38385.7

Estimates are models, not measurements. Accuracy falls off as reps rise and differs between lifters and exercises — treat the numbers as a starting point and adjust from how the set actually felt.

Enter the set you just did. The 1RM step uses the mean of Epley and Brzycki — the same pair Reppit's API uses for estimated rep maxes [S11] — and the projection table is solved as the exact inverse of that model, so a load taken from the table reproduces the same 1RM when entered back in.

How much to trust these numbers

Two separate sources of error stack here, and it is worth knowing which is which.

  1. Your RPE rating. Estimates of reps remaining are accurate to roughly one rep within about 0–5 reps of failure, and drift past two reps of error when 7–10 remain [3]. A misjudged RPE moves the whole calculation.
  2. The formula. Prediction equations correlate well with measured 1RM but carry exercise-specific bias — LeSuer and colleagues found the common equations over- or under-estimated differently across bench press, squat and deadlift — and they are most dependable at low rep counts [6].

The reassuring part: results are similar across a range of proximities to failure, so being one rep off is unlikely to change the training effect [4][5].

Reference: percentage of 1RM by reps and RPE

Generated from the same averaged Epley/Brzycki model the calculator uses, expressed as a percentage of estimated 1RM. Rounded to whole percent.

Approximate % of estimated 1RM for a set of N reps stopped at a given RPE. Generated from the same averaged Epley/Brzycki model as the calculator above, so the two agree — enter any load from this table into the calculator and it returns the 1RM it came from.
RepsRPE 10RPE 9RPE 8RPE 7RPE 6
1100%95%93%90%87%
295%93%90%87%85%
393%90%87%85%82%
587%85%82%80%77%
880%77%75%73%70%
1075%73%70%68%66%
1270%68%66%64%62%

How Reppit uses this

Reppit records an RPE on the set you just completed, on an integer 1–10 wheel, with each level described in reps left in reserve [8]. From completed sets it estimates 1RM through 12RM using Epley, Brzycki and their average, and refuses inputs of 37 reps or more because the Brzycki denominator breaks down there [7].

What does RPE 8 mean?

About two reps left in reserve — a hard set that is not a grind [1]. In the Reppit app, RPE 8 is labelled "Hard: about two reps left. Strong effort, not a grind yet." [8]

Why does my calculated 1RM differ between exercises?

Because rep-max formulas carry exercise-specific bias; the same equation can over-estimate one lift and under-estimate another [6]. Compare a lift against its own history rather than across lifts.

Is the percentage table exact?

No. It is a model, rounded to whole percent, and individual variation at a given rep count is substantial. Use it to choose a starting load and correct from the RPE of the set you actually do.

Can I use RPE 10 every session?

You can, but there is little reason to: outcomes across a range of proximities to failure are similar, and going to failure costs more fatigue for at most a trivial benefit [4][5].

Sources

  1. [1]Novel Resistance Training-Specific Rating of Perceived Exertion Scale Measuring Repetitions in Reserve Journal of Strength and Conditioning Research, 2016. primarySupports: The RIR-anchored RPE scale and its validation against movement velocity in experienced and novice squatters.
  2. [2]Application of the Repetitions in Reserve-Based Rating of Perceived Exertion Scale for Resistance Training Strength and Conditioning Journal, 2016. guidelineSupports: Practical application of the RIR-based RPE scale for prescribing and autoregulating resistance training.
  3. [3]Accuracy in Estimating Repetitions to Failure During Resistance Exercise Journal of Strength and Conditioning Research, 2017. primarySupports: Estimation error of roughly one repetition near failure, rising above two repetitions when many reps remain.
  4. [4]Influence of Resistance Training Proximity-to-Failure on Skeletal Muscle Hypertrophy: A Systematic Review with Meta-analysis Sports Medicine, 2023. meta-analysisSupports: At most a trivial hypertrophy advantage for training to failure versus stopping short of it.
  5. [5]Exploring the Dose-Response Relationship Between Estimated Resistance Training Proximity to Failure, Strength Gain, and Muscle Hypertrophy: A Series of Meta-Regressions Sports Medicine, 2024. meta-analysisSupports: Strength gain largely unaffected by proximity to failure, with a small hypertrophy dose-response favouring closer to failure.
  6. [6]The Accuracy of Prediction Equations for Estimating 1-RM Performance in the Bench Press, Squat, and Deadlift Journal of Strength and Conditioning Research, 1997. primarySupports: Accuracy and exercise-specific bias of 1RM prediction equations, including Epley and Brzycki.
  7. [7]Reppit estimated rep maxes: Epley, Brzycki and their average Reppit (shipped product behaviour), 2026. product-sourceSupports: Reppit's 1RM-12RM estimation from a completed set using Epley, Brzycki and their mean, with a 37-repetition limit.
  8. [8]Reppit workout logging: per-set RPE wheel and in-app RPE explainer Reppit (shipped product behaviour), 2026. product-sourceSupports: Reppit's integer 1-10 RPE wheel, its per-level wording, and the reps-in-reserve framing used in the app.