RPE and RIR: how to rate a set you just did

A log entry that reads 100 kg × 5 is missing the most important thing about the set: whether it was comfortable or nearly the last repetition you had.

Two scales exist to capture that, and they are the same scale.

The two scales

RIR — reps in reserve. How many more could you have done? Zero means none. Two means two.

RPE — rating of perceived exertion. A 1 to 10 scale where 10 means nothing left. Subtract from 10 to get RIR: RPE 8 is 2 RIR.

Most people find RIR easier, because "how many more could you have done" has a countable answer and "how hard was that out of ten" does not.

RPE	RIR	What it means
10	0	Could not have done another
9	1	One more was there
8	2	Two more
7	3	Three more
6	4+	Comfortably short

Why it belongs in the log

Consider two entries a week apart, both reading 100 kg × 5.

If the first was at 0 RIR and the second at 2, you got stronger. If the first was at 2 and the second at 0, you got weaker. Without the rating, both look like a flat line and you would conclude nothing happened.

That is a real signal, appearing in the period between load increases, and it is invisible without this field. It is why the rating is one of the four things worth recording in what to record in a training log, and what to skip.

Why most training sits at 1 to 3 RIR

Sets taken to absolute failure produce a strong stimulus and a large recovery cost. They fatigue the nervous system, extend how long the muscle needs before the next session, and degrade every set that follows in the same workout.

Sets stopped four or five repetitions short contribute relatively little.

The band between — one to three repetitions short — captures most of the stimulus at a fraction of the cost, which is why most well-designed programmes live there and use failure sparingly, typically on isolation work at the end of a session.

Calibrating your rating

Beginners are consistently wrong in one direction: they believe they are closer to failure than they are, often by three or four repetitions.

The fix is occasional calibration. On a safe exercise — a machine, a dumbbell curl, a leg press — take one set to genuine failure. Predict beforehand how many you will get. Compare.

Most people find they had considerably more than they thought. That correction is worth more to their training than any programme change, because it means every subsequent set has been closer to the productive range.

Using it to select load

RPE-based loading solves a real problem: a fixed percentage of your maximum is heavy on a bad day and light on a good one.

"Work up to a set of five at RPE 8" lets the load follow your actual capacity while the intent stays constant. On a day after poor sleep that might be 95 kg; on a good day 105. Both are the same training stimulus.

This is more flexible than percentage-based programming and requires the calibration above to work. For someone in their first six months, fixed loads with double progression are usually the better choice.

Its limits

It is subjective. Two people rating the same set will disagree, and so will the same person on different days. That is tolerable because you are comparing yourself with yourself.

It is harder on high-rep sets. Estimating reps in reserve on a set of twenty is much less reliable than on a set of five.

It drifts. Ratings tend to inflate over a training block as fatigue accumulates and a familiar weight starts to feel harder. That is information rather than error — it is often the first sign that a deload is due, per deload weeks: when you need one and how to tell.

In practice

Write a single digit after each working set. 100 × 5 @ 2. It takes a second and it is the difference between a log that records what the bar did and one that records what you did.

Reading those ratings across a block is one of the four progress signals in how to read real progress out of a training log, and it is the only one that shows improvement while the weight on the bar stays the same.