RPE stands for rating of perceived exertion. In weightlifting, programs commonly use a 1-10 scale tied to how many more reps you could have completed: RPE 10 means none, RPE 9 means about one, and RPE 8 means about two.
That lets you distinguish an easy 225x5 from a difficult 225x5. Log the rating with the set so you can compare effort as well as load over time.
The lifting RPE scale
Read "reps left" as reps you could have completed with acceptable technique.
| RPE | What the set felt like | Reps in reserve |
|---|---|---|
| 10 | No more reps were possible | 0 |
| 9.5 | No more reps, but slightly more load may have been possible | 0 |
| 9 | One more rep was possible | 1 |
| 8.5 | One more rep, possibly two | 1-2 |
| 8 | Two more reps were possible | 2 |
| 7.5 | Two more, possibly three | 2-3 |
| 7 | Three more reps were possible | 3 |
| 5-6 | Comfortable work with several reps left | 4+ |
| 1-4 | Very light work or warm-ups | Many |
Low-effort estimates are imprecise. You rarely need to distinguish an RPE 4 from a 4.5 to make a training decision. Nearer failure, half points can help describe the difference between sets.
The original Borg scale runs from 6 to 20 and is used for general exertion. A lifting program using reps-in-reserve anchors means a different scale, so check the program's definition before transferring a rating from running to lifting.
RPE and RIR describe the same set
RIR means reps in reserve. Near failure, the usual conversion is RPE = 10 minus RIR: two reps left corresponds to RPE 8, three left to RPE 7. At easier efforts, treat the categories more loosely.
A prescription of "five reps @8" asks for five reps with about two left. It does not prescribe a fixed weight. Choose a load from recent sessions and adjust after the warm-up if needed.
ChatRPE accepts explicit RPE or RIR in a log. "Two reps left" conveys more useful information than "hard." If you only know it was difficult, record that uncertainty rather than forcing a precise score.
What a rating is worth in weight
A rating on its own does not tell you what to load. The RPE chart is the bridge: find the reps you plan to do, read across to the effort you are aiming for, and it returns roughly what percentage of your one-rep max that combination costs.
| Reps | @ RPE 7 | @ RPE 8 | @ RPE 9 | @ RPE 10 |
|---|---|---|---|---|
| 1 | 89% | 92% | 96% | 100% |
| 2 | 86% | 89% | 92% | 96% |
| 3 | 84% | 86% | 89% | 92% |
| 4 | 81% | 84% | 86% | 89% |
| 5 | 79% | 81% | 84% | 86% |
| 6 | 76% | 79% | 81% | 84% |
| 7 | 74% | 76% | 79% | 81% |
| 8 | 72% | 74% | 76% | 79% |
| 10 | 67% | 69% | 72% | 74% |
Every cell runs the same arithmetic: add the reps you completed to the reps you left, then look up what that many reps to failure is worth. Five at RPE 8 is seven reps of capacity, so it lands on the same 81% as a set of seven taken to the limit.
Reading it forward picks a load. Your program says five at RPE 8 and your best squat is 400 lb, so 81% puts the opening guess near 325 lb. That is a guess to test in the warm-ups rather than a prescription. If 325 arrives at a 9, the day has told you something the chart could not.
Reading it backwards estimates a max, which is the part a rep calculator cannot manage. Squat 315 for five, call it an 8, divide by 0.81, and you get about 390 lb. The Epley formula reads that same set as 368, because weight and reps are all it has and it cannot see the two reps you kept. Count those reps first and the formula agrees with the chart.
These are the conventional percentages popularized by Mike Tuchscherer's Reactive Training Systems, rounded here to whole numbers. Published charts disagree by a point or two, which is smaller than the error in the rating itself. RTS recommends building your own chart from tested rep maxes and keeping a separate one per lift, since rep capacity at a given percentage is not the same for a deadlift as for a bench press. Treat the table as a starting point your own logged sets correct.
Rate the set you just completed
Ask how many more repetitions you could have done with similar technique. Breathlessness and discomfort can make a set feel harder without putting you as close to muscular failure as you think.
Accuracy is not uniform across the scale, and knowing where it degrades is what keeps the chart above honest. Ratings hold up best near failure and in lifters with experience: in the study that introduced the reps-in-reserve scale, experienced squatters rated a true one-rep max at 9.8 on average, while novices called that same maximal single a 9.0. The review that followed it advises against relying on the scale for lighter work several reps from failure, where the estimate is close to guesswork.
That is the practical reason a working-set rating carries weight and an RPE 4 on a warm-up does not. Consistent logging helps you compare your estimates with later performance. Bar speed can provide context, but speed alone is not a precise RPE measurement.
Record the rating promptly. A note such as "last rep slowed; another two looked possible" is useful even if you are unsure whether the set was a 7.5 or an 8. Use safe setups and coaching when learning your limits rather than testing failure on an unfamiliar heavy lift.
Use effort to adjust a load
If 185x8 was RPE 9 several weeks ago and is repeatedly around RPE 7 now, the same work is becoming easier. If effort rises across several comparable sessions, investigate recovery and the training dose before automatically adding weight.
The deload guide covers those recovery signals. The progressive overload guide explains how to decide between adding load, adding reps, or repeating a prescription.
Many strength programs use working sets around RPE 7-9, with the appropriate effort depending on the exercise and purpose. Follow the program's instructions rather than treating every set as a test. The powerlifting guide applies effort targets across a meet cycle.
Keep effort attached to the set
"Squats 275 lb for 3x5. First two sets around RPE 7, last set 8.5 because my brace became difficult to hold. Save it and compare with my previous squat sessions."
ChatRPE keeps the reported effort alongside weight and reps. Ask for matched-load comparisons or an effort chart, specifying the lift and period. A session-level note alone cannot tell you which set the rating belongs to.
The workout logging guide covers recording and correcting those entries. To use the ratings for today's prescription, see RPE-based training in ChatRPE.
Frequently asked questions
What does RPE mean in weightlifting and powerlifting?
Rating of perceived exertion. Lifting programs commonly anchor it to reps in reserve: the number describes how close the set came to failure with acceptable technique.
What does RPE 8 mean?
About two reps left. After five reps at RPE 8, you estimate that seven total reps would have been possible with that weight and similar technique.
What is the difference between RPE 9 and RPE 10?
RPE 9 means about one rep remained. RPE 10 means no further rep was available. Neither is a requirement for every working set.
What percentage of my max is five reps at RPE 8?
About 81%. The percentage is read off the reps you were capable of, not the reps you performed, so a five at RPE 8 is priced as a set of seven. Against a 400 lb squat that is near 325 lb, which is an opening load to check in the warm-ups rather than a number to trust blind.
Is RPE better than percentage-based training?
They serve different purposes and can be combined: percentages give a planned load, and RPE adjusts it for the day. The trial that matched the two directly, an eight-week comparison in trained men, found both groups added squat and bench strength, with the RPE group gaining somewhat more. That difference was not statistically significant across 21 lifters, so read it as no penalty for training by effort rather than proof of an advantage.
How precise does my rating need to be?
Use the best estimate you can support and record uncertainty. Repeated comparable sets are more informative than an isolated decimal rating, and the numbers you should work hardest to get right are the ones on heavy sets, where the scale is most reliable anyway.