Percentages prescribe weight from a past max. RPE prescribes weight from today's effort. Both work, both fail in specific ways, and the head-to-head research is closer than either camp admits. The smart answer for most lifters is a hybrid.
If RPE is new to you, read RPE and RIR explained first. This article assumes you know the scale and answers the next question: should your working weights come from a percentage table or from your own effort rating?
How Each System Assigns Weight
Percentage-based training starts from your 1RM and prescribes loads as fractions of it: 5 reps at 75 percent, 3 at 85 percent. The program decides everything in advance. Your job is to execute.
RPE-based training prescribes a rep count and a target effort: 5 reps at RPE 8, meaning a weight you could lift for about two more reps. You pick the load each session based on how the warm-up sets move.
One system trusts the plan. The other trusts your judgment. The whole debate is about which one is wrong less often.
The Conversion Chart
The two systems describe the same thing from different ends, which is why a conversion chart exists. This one is adapted from the table published by Helms and colleagues (2016) in the Strength and Conditioning Journal, built on Mike Tuchscherer's Reactive Training Systems chart:
| Reps | RPE 10 | RPE 9 | RPE 8 | RPE 7 |
|---|---|---|---|---|
| 1 | 100% | 95% | 91% | 87% |
| 2 | 95% | 91% | 87% | 85% |
| 3 | 91% | 87% | 85% | 83% |
| 4 | 87% | 85% | 83% | 81% |
| 5 | 85% | 83% | 81% | 79% |
| 6 | 83% | 81% | 79% | 76% |
| 7 | 81% | 79% | 76% | 73% |
| 8 | 79% | 76% | 73% | 70% |
Read it either way: 5 reps at RPE 8 sits around 81 percent of 1RM, and 81 percent for 5 should feel like an RPE 8.
One honest caveat the chart's own authors attach: only a few cells come from measured data, the rest are estimates, and individuals vary. Treat it as a translation guide, not a law.
Where Percentages Fail
Your 1RM is probably stale. A percentage is only as good as the max it comes from. Helms and colleagues point out that the reference max often reflects a previous performance that no longer represents you: suppressed by fatigue when you tested it, outgrown two months later, or never tested at all and just estimated.
Your strength is not the same every day. Estimates from velocity-based training research suggest daily maximal strength can swing well over 10 percent in either direction depending on sleep, stress, and accumulated fatigue. A fixed 85 percent is a different set on a good Friday than after a bad week.
The same percentage is a different effort for different lifters. In the data behind the chart above, lifters doing 8 reps at 70 percent reported anywhere from about 1 to 5 reps in reserve. Identical prescription, wildly different stimulus.
The percentage world quietly admits all this. Jim Wendler's 5/3/1 runs every percentage off a training max of about 90 percent of your real max, with later versions recommending 85 percent. That buffer exists precisely because fixed percentages on a true max break on bad days. See the 5/3/1 for beginners guide.
Where RPE Fails
Lifters misjudge their own effort, in a predictable direction. A 2022 meta-analysis by Halperin and colleagues pooled 12 studies and found lifters underpredict their reps in reserve by about one rep on average. When you call a set RPE 8, it is often really an RPE 7. You are probably sandbagging, not overreaching.
The error grows far from failure. Hackett and colleagues (2017) found prediction error under one rep when lifters were within about three reps of failure, but more than two reps of error when they were 7 to 10 reps out. RPE 9 is a much more trustworthy report than RPE 6.
Beginners misread maximal effort itself. In the study that established the modern scale, Zourdos and colleagues (2016) had novice and experienced squatters rate a true 1RM attempt. The experienced group called it 9.8. The novices called their actual max an RPE 9, on average, believing they had a rep left at a weight they could not have added a rep to. Whether accuracy reliably improves with experience is less settled than commonly claimed, but accuracy clearly improves close to failure and on heavier loads for everyone.
What the Head-to-Head Studies Show
- Graham and Cleather (2021): 31 trained men, 12 weeks, matched programs. The RIR-autoregulated group gained more: front squat up 11.7 percent vs 8.3, back squat up 10.8 vs 7.1. The strongest result in RPE's favor.
- Helms and colleagues (2018): 8 weeks of squat and bench in trained lifters. Both groups improved, differences not statistically significant, with a slight trend favoring the RPE group.
- Hickmott and colleagues (2022), meta-analysis: across 15 studies and 441 participants, autoregulated loading did not significantly beat fixed percentages, with a near-significant trend favoring RPE-style methods.
Fair verdict: the evidence leans slightly toward RPE for trained lifters and proves nothing decisive. Nobody should switch systems expecting a different body. The real differences are practical.
The Hybrid Most Lifters Should Run
The systems fix each other's weaknesses, which is why most good modern programs already blend them:
- Percentages set the skeleton. Main lifts get percentage-based targets so progression is planned and the log stays comparable week to week.
- RPE caps the bad days. Add a ceiling to every prescribed set: "75 percent for 5, but stop the set if it hits RPE 9." On a bad day the cap saves the session. On a good day it never activates.
- RPE runs the accessories. Curls and lateral raises do not need a tested 1RM. "3 sets of 10 at RPE 8" plus double progression covers every accessory in your program.
- AMRAP sets calibrate the percentages. A rep-out set tells you your current estimated 1RM. Run it through a 1RM calculator and your next block's percentages come from present-day strength, not a six-month-old max.
How to Calibrate Your RPE
Helms and colleagues recommend exactly what the accuracy research implies: practice before you trust it.
- Log a predicted RIR on your last working set of each lift while still running your normal program. Do not change any weights yet.
- Occasionally test it. Once every few weeks, take that final set to actual failure on a safe exercise and compare reality to your prediction. See training to failure for where that is appropriate.
- Trust your ratings at RPE 8 and above first. That is where everyone is most accurate. Far-from-failure ratings stay rough estimates for a long time.
Which Should You Use?
- First year of lifting: percentages or simple linear progression. Your RPE reports are not reliable yet, and a beginner program already manages effort for you. Log RIR for practice.
- Intermediate, training hard, life is variable: the hybrid above. Percentage skeleton, RPE cap, RPE accessories.
- Advanced, competing, or coaching yourself: RPE earns its keep here. Daily fluctuations are a bigger fraction of your progress, and your ratings are calibrated by thousands of logged sets.
Looking for a workout tracker?
If you want to make real progress and build discipline in the gym, use Virtus Athlete. Free on iOS and Android.
