Progression

RPE vs RIR: two names for one question, and the one to log

You finish a set of eight, rack the bar, and someone asks how hard it was. You could say "RPE 8". You could say "two in the tank". You've just answered the same question twice, in two dialects, and half the internet's arguments about RPE versus RIR are about which dialect is better.

That's not the interesting part. The interesting part is that whichever one you say, you're probably wrong by about a rep. And which way you're wrong matters more than which scale you used to be wrong in.

The short version

• RIR is reps in reserve: how many more you could have done.
• Lifting RPE is the same idea on a 1 to 10 score. RPE 10 is 0 RIR, RPE 9 is 1, RPE 8 is 2.
• On average, people underestimate how many reps they have left by about one.
• Estimates are better on heavier sets, closer to the end of the set, and in later sets.
• Log RIR as a number next to your reps. It's the more concrete question, and it adds straight onto your rep count.

They're the same scale pointing in opposite directions

The version of RPE most lifters use comes from a scale tested by Zourdos and colleagues, who anchored each RPE value to reps in reserve: RPE 10 meant zero reps left, RPE 9 meant one, and so on down. That anchoring is what made the scale usable for lifting: a number out of ten became a claim about reps you could check.

So the conversion is not an approximation. On that scale, RPE and RIR are the same measurement written two ways. RPE 7 is 3 RIR. RPE 8.5 is "somewhere between one and two left, and I'm not sure".

The difference is in what the question feels like, not what it measures.

RPE asks you to rate an effort. RIR asks you to count something you didn't do. Counting is easier than rating for most people, which is why RIR tends to be the better place to start.

Why RIR is easier to log honestly

Three practical reasons, none of them deep.

First, RIR is a whole number about reps, and you're already writing reps down. 8 reps with 2 left sits naturally in a log as 8 @2. An RPE number sits next to it as a second, different kind of quantity.

Second, it adds. Eight reps with two in reserve is ten reps of capacity, and that sum is what makes two sessions comparable when the reps and the effort both changed. We use exactly that: effective reps are reps plus RIR, so 7 reps @1 is the same day as 8 reps @0. The explanation lives in what an effective rep is.

Third, it fails more obviously. Write 8 reps with 3 in reserve, then manage only 9 with nothing left at the same weight next week, and the log shows a drop from 11 to 9 that you can go and question. Either you got weaker, or last week's 3 was really 1. An RPE number that drifts from 7.5 to 8 doesn't announce itself the same way.

None of that makes RPE wrong. Lots of powerlifting coaches use it well. It's just one translation step further from the thing being measured.

The real problem: almost everyone is off by about a rep

Here's the research that matters more than the naming debate. Halperin and colleagues pooled 12 studies covering 414 participants who predicted how many reps they had left and then kept going to failure to check. On average, people underpredicted by about 0.95 reps.

Underpredicted means they had more left than they said. Someone writing 1 RIR typically had about 2.

That average hides a lot. Heterogeneity between studies was very high, so in some settings people were close and in others they were far off. But the patterns inside the data are useful:

• Heavier sets were judged better. Accuracy was much worse on sets that went past 12 reps.
• Predictions made nearer the end of a set were slightly more accurate than ones made at the start.
• Later sets were judged slightly better than first sets.
• Training experience did not seem to improve accuracy in the pooled data.

That last one surprises people, and it should change how you read your own numbers. Years of training don't automatically calibrate you. Checking does.

How most lifters misjudge zero

Zero is an easy number to misjudge. "To failure" in a training log can mean "it got really uncomfortable", and the average error in the research above suggests that feeling can arrive with a rep or so still available.

That has a direct cost when you're comparing sessions. Say last week you logged 8 reps at 0 RIR, but there was secretly one more rep in you. This week, no stronger than before, you take the set to true failure and get 9. The log shows progress that didn't happen. The error doesn't stay in one set. It leaks into every comparison that set is part of.

There's also the opposite problem: logging nothing. If you don't enter an RIR at all, the honest assumption is zero, which treats every unlabelled set as a true failure set. That's the safest default for comparing like with like, but it does mean the number is only as good as the effort you put into recording it.

Your RIR is a measurement, not a mood. Treat it like the weight on the bar: write what was true, not what felt dramatic.

How to get better at it

One practical way to check yourself is calibration. Occasionally take a set you'd normally stop at 2 RIR and keep going, safely, on a lift where failure won't hurt you: a machine, a cable, a dumbbell curl. See whether 2 was really 2.

Do that a few times and you'll get a feel for how far off your own estimates run. The pooled research gives you a starting assumption, roughly a rep conservative on average, and in that data the error did not differ much between people or between upper and lower body lifts. Checking your own sets now and then is what keeps that assumption honest, rather than something you take on faith.

Two restraints worth keeping:

• Don't do calibration sets on a heavy barbell squat or bench without a spotter or safeties.
• Don't change how you record RIR halfway through a training block. A sudden shift in honesty looks exactly like a sudden shift in strength.

If you're deciding how often to actually go to zero on purpose, how often you should train to failure covers that separately.

So which should you use?

Whichever you'll write down every time. If you're starting fresh, RIR: it's the more concrete question, it sits next to reps in the same unit, and it adds into a single comparable number.

If you already use RPE and your coach programs in it, there's no reason to switch. Convert in your head and keep going.

The honest concession is that neither scale fixes the underlying problem. Both are guesses, and the evidence says the guesses run about a rep conservative on average. A logged guess you check now and then is still far more useful than a perfect scale you never fill in.

The scale is a detail. The habit of writing it honestly is the whole thing.

Common questions

What is the difference between RPE and RIR?

RIR is the number of reps you think you had left at the end of a set. RPE, as lifters use it, is a 1 to 10 effort score built on that same idea, where 10 means nothing left and 9 means one rep left. One counts down, the other counts up.

Is RPE 8 the same as 2 RIR?

On the lifting version of the scale, yes. RPE 10 is 0 reps in reserve, RPE 9 is 1, RPE 8 is 2. Half points like 8.5 mean you were not sure whether one or two reps were left.

How accurate is RIR?

Not very, on average. A meta-analysis of 12 studies found people underestimated how many reps they had left by about one rep, with large differences between studies. Predictions got better with fewer reps per set, closer to the end of the set, and in later sets.

Should beginners use RPE or RIR?

RIR is easier to start with because it asks a concrete question about reps rather than an abstract score. Beginners will be inaccurate either way, and the research suggests experience alone does not fix that, so treat early numbers as rough and get better by checking them.