RPE measures the reps you left, not how hard it felt.

Rating of Perceived Exertion isn’t a mood ring. As I use it, and as it’s used in every serious strength setting, it answers one question: how many more reps could you have completed with that load before your form came apart? An 8 means two. A 9 means one. A 10 means none. That’s the whole instrument.

The scale is a rep count wearing a different hat

• 10 — nothing left. The last rep slowed and you finished it anyway.

• 9 — one more, and you’d have had to fight for it.

• 8 — two more, with bar speed still even across reps.

• 7 — three more. This is working weight, not a test.

• 6 and below — warm-ups, technique, speed work. You should never grind here.

So when a session reads 4×5 @8 RPE · 180s · 30X1, the instruction is: pick a load where the fifth rep of each set leaves two in reserve, rest three minutes, take three seconds to lower the bar. If set one comes in at 6, the weight was light and you add to it for set two. That isn’t cheating the program. That is the program.

Why I don’t hand you percentages instead

A percentage is anchored to a one-rep max you tested on one specific Tuesday. Your actual capacity moves ten to twenty percent around that number depending on sleep, illness, how much you walked at the weekend, and whether you spent the morning in a meeting you’d rather not have had. A percentage-based program written in March is a guess about your June. RPE re-anchors the guess every single set.

Percentages still have a job. In HEAVY I use them for warm-up ramps and for the first week of a new block, because they take decisions away while you’re learning a movement. After that the RPE governs and the percentage is a suggestion.

Calibrating yours takes about three weeks

Most people are wrong by one to two points when they start, and almost always in the same direction — they call a 7 a 9. Three habits fix it.

• Predict before you rack. Say the number out loud before the bar goes back, then write it down. Guessing after you’ve caught your breath is a different and much easier question.

• Film one working set per lift, from the side, at hip height. Bar speed on the last rep tells you more than your memory will. If rep five moved at the same speed as rep two, that wasn’t a 9.

• Once a week, on something you can fail safely — a machine, a dumbbell press, a leg press — take a set until the next rep genuinely won’t happen. Count how far past your guess it went. That gap is your calibration error, and it shrinks fast once you’ve measured it.

The two ways people get it wrong

Sandbagging is the common one and it’s quiet. You train at a true 6 for a whole block, everything feels fine, nothing changes, and you conclude the program didn’t work. It did work. You never ran it. If your log is a flat line of the same weights for four weeks and every session felt comfortable, you have an RPE problem, not a programming problem.

Inflating is rarer and costs more. Everything reads 9, every set is a grind, and the fatigue you’re accumulating exceeds what the block was designed around. You’ll get away with it for two or three weeks. Then warm-ups start taking longer, sleep gets patchy, and a joint you’ve never thought about starts commenting.

Once your numbers are honest, everything downstream gets simpler. Load selection stops being a negotiation, deloads land where they were meant to, and I can read your log from six hundred miles away and know what your week was actually like.