RIR vs RPE: Which Should You Use to Autoregulate Your Training?

Short answer: it doesn't matter much, because RIR and RPE are the same thing wearing two different jerseys. On the modern resistance-training scale, RPE 8 means 2 reps in reserve, RPE 9 means 1, RPE 10 means you couldn't do another rep if your life depended on it. Pick whichever notation you find easier to think in. The real RIR vs RPE question — the one almost nobody asks — isn't "which scale" but "can you actually feel where failure is?" Spoiler: you mostly can't, and that's where this gets interesting.

RIR (reps in reserve) is your estimate of how many more reps you could have done in a set before hitting true muscle failure. RPE (rating of perceived exertion) on the lifting-specific scale is the same estimate flipped into a 1-10 effort number. Two faces, one idea: autoregulation — adjusting today's load to today's body instead of obeying a rigid percentage.

I've coached this stuff for years, and I want to give you the conversion, the goal-specific recommendation, and then the uncomfortable part both scales share — plus how I actually fix it in practice.

What's the difference between RIR and RPE?

Functionally, almost nothing. On the RIR-based RPE scale that the strength world adopted after Zourdos et al. (2016), RPE is just RIR inverted into a 1-10 score. RPE 10 = 0 RIR. RPE 7 = 3 RIR. The one genuine difference: RPE adds half-point increments. RPE 9.5 means "I couldn't add another rep, but I could've added load" — a near-maximal single state that whole-number RIR literally cannot express. That's it. That's the whole difference.

So when someone online frames this as RIR vs RPE like they're rival systems, they've misunderstood the scale. They aren't competing. RPE is RIR with one extra decimal place for heavy singles.

RIR vs RPE conversion (table)

Here's the conversion, straight from the Zourdos scale (Table 2 of the original validation study, popularized in the excellent Ripped Body RPE guide):

Print it, tape it to your phone, whatever. Notice how the bottom of the table gets vague fast — "4 to 6 reps left" is a wide range, and that vagueness isn't laziness in the table. It's honest. Your perception genuinely gets worse the farther you are from failure, which brings us to the part that matters.

Which is better for hypertrophy vs strength?

If I'm forced to split them by goal: RPE wins for strength and powerlifting, RIR wins for high-rep hypertrophy work. RPE's half-points (9.5) exist precisely to encode load-limited singles near your 1RM — exactly the territory powerlifters live in. For a 5x5 bench at RPE 8, that half-point resolution earns its keep. For bodybuilding, where you're chasing a pump in the 10-15 rep range, "leave 2 RIR" is more intuitive than translating to an RPE number, and the half-points are useless anyway.

Does autoregulation itself beat fixed percentages? Yes, and that's settled. A 2025 network meta-analysis of 19 RCTs (438 participants) found every autoregulated method — RPE, velocity-based, and APRE — outperformed percentage-based training for 1RM gains. RPE/RIR ranked second for squat. Why? A fixed 80% 1RM is a brutal grind on a bad-sleep day and a warm-up on a great one. RPE/RIR auto-adjusts to your daily readiness. Percentages can't. So whichever notation you choose, you're already ahead of the guy chained to a spreadsheet of percentages.

The weakness both share — and how to fix it

Here's the catch nobody markets: both scales depend entirely on your self-perception, and your self-perception is bad. Not "could be better" bad. Genuinely unreliable bad, especially where you use it most.

The research is brutal and consistent. Across common exercises, the standard error of predicting reps to failure runs 2.64 to 3.38 reps — leg press 3.38, chest press 2.91, pulldown 2.95 (study here). Translation: when you call a set "2 RIR," true failure was routinely 4-5 reps away. And the error isn't random. Trained lifters systematically underpredict — they stop with gas in the tank. In a Frontiers in Psychology study, people who thought they hit their rep max actually had about 2 more reps in them; their "failure" averaged 11.6-13.3 reps when true failure was 14.1-14.3.

It gets worse the farther from failure you go. In a 70% 1RM squat set averaging ~16 reps, experienced lifters were off by 5.15 ± 2.92 reps when rating their position relative to failure (MASS Research Review). Five reps. That's not a rounding error — that's a different training stimulus entirely.

The cruel irony: accuracy is good only at 0-2 RIR on heavy, low-rep sets. It falls apart at 3+ RIR and 12+ reps — which is exactly the high-rep hypertrophy zone where lifters lean on autoregulation hardest. Experience helps (experts land within ~1-2 reps, beginners are off by 4-5), but Zourdos's 2019 follow-up showed even at 1 RIR, trained lifters stay imperfect. Proximity helps. It doesn't cure the problem.

So how do you fix a guess? You anchor it to something objective.

The original Zourdos validation wasn't just a feelings survey — it correlated RPE against average concentric bar velocity at r = -0.88 in experienced squatters. As you fatigue, reps physically slow down. That velocity decay is the external, measurable truth that the subjective scale was trying to approximate all along. It's the basis for velocity-based training, and it's why a barbell encoder beats a guess.

This is exactly the gap Riven is built for. Riven reads rep-speed decay straight off your Apple Watch — no barbell clip, no camera, no extra hardware — and converts it into a 0-100 failure-proximity score per set, with heart rate as context. It's most useful precisely where your perception is worst: high-rep, far-from-failure sets where you'd otherwise be 4-5 reps off. Instead of guessing "eh, 2 RIR," you get an objective read on whether you actually approached failure.

I'll be straight with you, because this audience smells marketing instantly: the wrist signal is a proxy. It reads roughly half the velocity loss a $300 linear position transducer sees at equal fatigue, so it's calibrated per exercise category, not read as raw bar speed. Don't expect lab-grade EMG numbers. Do expect an objective on-wrist anchor that beats guessing — and that quietly trains your own perception faster than the 3+ years it took the experts in those studies to get accurate.

That's the honest pitch. RIR and RPE both work, autoregulation beats fixed percentages, and the bottleneck isn't your scale — it's your self-perception. Sharpen that with an objective signal and either notation gets a lot more trustworthy.

FAQ

What's the difference between RIR and RPE?

On the modern lifting scale, almost none — they're the same construct. RPE is RIR inverted into a 1-10 number (RPE 8 = 2 RIR). The only real difference is RPE's half-points (e.g. 9.5) for load-limited near-maximal singles, which whole-rep RIR can't express.

Is RIR more objective than RPE because it counts reps?

No. Both rely on the identical internal perception of effort. "Reps left" is itself an estimate, not a measurement — counting it doesn't make it objective. Neither scale is objective; both are subjective.

If I leave 2 RIR, am I really 2 reps from failure?

Usually not. Self-perception error averages 2.6-3.4 reps, and trained lifters tend to undershoot — so "2 RIR" often means true failure was 4-5 reps away. That undershoot silently reduces hypertrophy stimulus on high-RIR sets.

Is percentage-based loading more precise than RPE/RIR?

No. A fixed %1RM ignores daily readiness — it's a different effort on a tired day than a fresh one. In meta-analytic evidence, autoregulated RPE/RIR matched or beat percentage-based training for both strength and hypertrophy.

Can I get more accurate at RIR/RPE?

Yes — it's a trainable skill, not innate. Experts (>3 years) land within ~1-2 reps; beginners are off by 4-5. Practicing near failure helps, and an objective anchor like measured velocity loss accelerates building that perceptual skill.

Sources

• Zourdos et al. (2016) — Novel Resistance Training-Specific RPE Scale Measuring Repetitions in Reserve, J Strength Cond Res
• "Just One More Rep!" — Ability to Predict Proximity to Task Failure in Resistance Trained Persons (Frontiers in Psychology, 2020)
• Ability to Predict Repetitions to Momentary Failure Is Not Perfectly Accurate (PMC5712461)
• Zourdos (2019) — Proximity to Failure and Total Repetitions Influence Accuracy of Intraset RIR-Based RPE (PubMed)
• Autoregulated Resistance Training for Maximal Strength: Network Meta-Analysis (PMC12336695)
• RPE and RIR: The Complete Guide — MASS Research Review
• A Guide To Using RPE and RIR In Your Training — Ripped Body