Training by RPE: how to stop guessing at percentages

A fixed percentage ignores whether you slept. Reps in reserve lets the load follow your actual capacity — though most people stop a rep early.

Your programme says five sets of five at eighty percent. You slept four hours, you have had a stressful week, and the third set moves like a car. Do you finish the prescription as written?

Percentage-based programming has no answer to that question, because it was set weeks ago by someone who could not know how today would go. RPE was designed to answer it.

What RPE means in lifting

The lifting version of RPE is not the ten-point exertion scale used in cardio. It is defined by reps in reserve — how many more repetitions you could have completed.

RPE 10 means none left: a true maximal effort. RPE 9 means one left, RPE 8 means two, and so on. The two numbers are simply inverses of each other, and the conversion is RPE = 10 − RIR.

This definition is what makes the scale useful. “How hard did that feel” is subjective and drifts with mood. “How many more could you have done” is a question about the set itself, and people are considerably better at answering it.

Why fixed percentages fail

A percentage of your one-rep max is a fixed prescription applied to a body whose capacity changes daily.

Sleep, stress, nutrition, accumulated fatigue and even time of day all move what you can do on a given afternoon. Eighty percent when fresh might be a comfortable five reps. The same eighty percent after a poor week might be a grinding three. Written as a percentage, those are the same instruction; experienced in the gym, they are entirely different sessions.

There is a second problem. The percentage is derived from a one-rep max that is itself an estimate, often weeks old, and frequently produced by a formula that disagrees with four other formulas. You are applying precise arithmetic to an imprecise input.

Autoregulation is not a softer approach. It is a more honest one — the load is set by what you can actually do today rather than by what you could do last month.

How well can people actually judge it

The obvious objection is that RPE is subjective, so perhaps it is no better than guessing.

The research is more encouraging than that. Work validating the RIR-based scale found that resistance-trained individuals estimate proximity to failure reasonably well, and accuracy improves as the set gets closer to failure — people are better at knowing they have one rep left than at knowing they have four.

The consistent finding is the direction of the error, and it applies to almost everyone. Pooling twelve studies and 414 participants, people under-predict their remaining repetitions by about one — they believe themselves closer to failure than they are. ³ If you stop at what feels like two in reserve, you were probably closer to three.

Whether experience fixes this is less clear than it is usually presented. The study that introduced the scale found trained lifters more accurate than novices, but the meta-analysis found no overall effect of training experience. ¹ ³ Treat “novices are worse at this” as plausible and unproven rather than established.

One caveat is solid: accuracy declines at high rep counts, where the sensation of effort becomes dominated by breathing and discomfort rather than by proximity to mechanical failure.

Whether this actually beats a fixed percentage plan is less settled than enthusiasts suggest. A 12-week trial comparing reps-in-reserve autoregulation against fixed loading found greater strength gains for the autoregulated group. ⁴ But a systematic review pooling fifteen studies concluded that autoregulated and standardised load prescription produce similar strength improvements, while a separate meta-analysis found autoregulation ahead. The meta-analyses disagree with each other.

The honest summary is that reps in reserve is at least as good as percentage-based loading and possibly better, with the evidence not yet strong enough to say which. That is still a reasonable argument for using it — it costs nothing, it adapts to bad days, and the downside case is parity — but it is a weaker claim than “autoregulation is better”, which is what gets repeated.

Using it in a programme

Put most volume work at RPE 7–8. Two or three reps in reserve accumulates productive work without generating fatigue you have to recover from for days.

Reserve RPE 9–10 for rare occasions. Testing, a peaking block, or the last set of a final week. Sets taken to genuine failure are disproportionately fatiguing relative to the extra stimulus they provide.

Use it as a readiness signal. If your usual RPE 8 weight suddenly feels like RPE 10, that is real information about recovery. Two consecutive sessions of that pattern is a reasonable prompt to reduce volume for a week.

Combine, do not replace. Many good programmes use percentages as a planning scaffold and RPE as a daily correction: the plan says roughly this load, the RPE cap says do not exceed this effort. Our converter goes both directions, so you can find the load for a target RPE or work out what RPE a set you already did represented.

The bottom line

RPE will not make a badly designed programme good. What it does is stop a reasonable programme from becoming unreasonable on the days your body did not read the spreadsheet.

If you are new to it, expect to be wrong at first — and expect to be wrong in the direction of stopping too early. Record the RPE you thought a set was and how many reps you actually had left when you occasionally push one to failure. The gap closes faster than you would guess.