You have to write one of them in the intensity column, and both options describe the same set. RPE 8 and 2 RIR are not close to each other, they are identical — two reps short of failure, described from opposite ends.
So the choice is not about which scale is more accurate or better supported. It is about which question your client can answer honestly at the end of a hard set, when they are tired and want to go home. That is a coaching question, not a science one, and the answer is different for a six-month client and a lifter of ten years.
Write RIR for almost everyone.
Write RPE only when the client is experienced enough that the half-points do work.
Never run both in the same program.
RIR asks for a fact, RPE asks for a judgement
This is the whole difference, and it is bigger in practice than it looks on paper.
"How many more reps could you have done?" is a question about a thing that either was or was not true. The client counts. They may count wrong, but they are estimating a quantity they have direct experience of.
"Rate that set out of ten" is a different cognitive job. The client has to hold a ten-point scale in their head, remember what each number is anchored to, and map a feeling onto it. Most clients cannot do that, so they do something else instead — they rate how unpleasant the set was. That is why you get RPE 9 on a set of twenty leg extensions and RPE 7 on a heavy triple that visibly ground to a halt. Both ratings are honest. Neither is measuring proximity to failure.
A client whose ratings never move while the load never moves. Four weeks of RPE 8 at the same weight is not a lifter holding intensity steady. It is a lifter reporting "that was hard", which was true every week and told you nothing.
Switching that client to RIR usually fixes it inside a fortnight, without any conversation about scales. The question changed, so the answer changed.
When RPE earns its place
None of that makes RPE the worse scale. It makes it the scale with a prerequisite.
For an experienced lifter — someone with years of heavy training, a real reference for failure and consistent technique — RPE is the better instrument for one reason: the half-points. RPE 8.5 and 9.5 are useful positions, and RIR has no way to express them. You cannot write 1.5 reps in reserve without it reading as a typo.
On a heavy single or double, the difference between "one rep left" and "one rep left, maybe two" is a real difference in load, often five to ten kilos on a squat. An advanced lifter can feel it and can report it. Everyone else cannot, and asking them to try adds a decimal place of noise to a number that was already rough. A client who cannot reliably tell 2 RIR from 3 RIR is not going to give you a meaningful 8.5.
So the rule is straightforward. Half-points are precision for lifters who can produce them and false precision for everyone else. If you would not trust a client's whole numbers, do not give them halves.
| Write RIR | Write RPE | |
|---|---|---|
| Client experience | Under about three years of consistent lifting | Advanced, with a real reference for failure |
| Question they answer | "How many more could you have done?" | "How close to failure was that, to the half point?" |
| Resolution | Whole reps | Half points on heavy low-rep work |
| Failure risk | Under-reports reserve early on | Rates unpleasantness instead of proximity |
| Best on | Everything, but especially accessories | Heavy compounds at low reps |
| Time to usable data | Two to four weeks | Two to three months |
Pick one and keep it for the whole program
The most common mistake is not choosing the wrong scale. It is choosing both.
A program that prescribes RPE 8 on the main lift and 2 RIR on the accessories is asking the client to run two mental conversions per session and to remember which column belongs to which exercise. They will do it wrong. Usually they read RIR as RPE somewhere in week two, take a set of curls to a genuine 8 reps past where you wanted them, and you find out three weeks later when the elbow hurts.
Mixing also destroys the thing you collect ratings for. You want to compare this week's effort to last week's on the same movement. That comparison is only clean when the units are the same all the way down the block.
There is one defensible exception: an advanced client on a peaking block where the competition lifts run on RPE and the general preparation work runs on RIR, and the client has explicitly been taught both. That is a small number of clients. If you are not certain yours is one, use one scale.
Whichever you pick, the rating has to survive the week. In Fitsly the intensity scale is a per-set tracking column on the exercise, so RPE or RIR sits next to weight and reps in the training log and comes back attached to the set rather than in a message you have to scroll for. The RPE column takes 0 to 10 in steps of 0.5; RIR takes whole numbers 0 to 10. Set the default on the exercise and it carries into every workout that uses it.
The client who cannot rate either
Some clients report nothing usable. They give you the same number every set, or a number that contradicts the reps they logged, or nothing at all. This is common and it is not a character flaw — they have never taken a set to failure, so every rating is a guess against a reference they do not have.
Do not keep asking. Three steps, in order:
Give them the reference. One set to genuine failure on something safe — leg press, machine row, a dumbbell press with the dumbbells over a bench. Not a back squat, not a barbell bench without a spotter. Until a client has felt a real zero, the rest of the scale is imaginary.
Prescribe a range. 2–3 RIR rather than 2 RIR. A client who is half a rep out is still training where you wanted them, and the range removes the pressure to produce a precise number they do not have.
Take the rating out of the program entirely and control load yourself. Fixed weights, a set progression, and rep targets you adjust from the logged reps.
Step three is not a failure. For a beginner it is the correct method, and it is the reason training volume is a more useful lever than intensity in someone's first year. Come back to RIR at six months.
Percentages with an RPE cap
There is a third option that gets treated as a compromise and is actually the right answer for a specific case: prescribe the load as a percentage and add an effort ceiling.
Written out, a line looks like this: 5 × 3 at 82.5%, cap RPE 9. The client loads the calculated weight and does the work. If a set hits the cap before the reps are done, they stop and drop the next set by about five per cent.
This suits a client with a recent, honestly tested one-rep max who is running a block where the absolute loads matter — a strength peak, a return from a layoff, a powerlifter eight weeks out. The percentage does the programming and the cap catches the bad day, which is the failure mode a pure percentage block has no answer for.
It is the wrong tool for general hypertrophy work, where the max is stale or was never tested and the cap ends up doing all the work anyway. If the cap is what determines the load most weeks, delete the percentage and prescribe the effort.
Heavy compounds and high-rep accessories are different jobs
The scale should also change with what the set is doing, and this is where most programs are sloppiest.
On heavy compounds at one to five reps, ratings are at their most accurate. The client is close to failure, bar speed is obviously falling, and the gap between "one left" and "two left" is visible. This is the only place half-point RPE is worth the trouble, and it is where a rating genuinely changes what you write next week.
On high-rep accessory work — sets of fifteen to twenty on a lateral raise, a leg extension, a cable row — ratings get worse as reps go up. The burn arrives long before failure does, so a client finishing a set of eighteen will tell you they had nothing left when four reps were there. Prescribe 0–2 RIR on that work and accept that you are setting a direction rather than measuring a position. Do not bother asking for a rating back on every set of an isolation exercise; you will not act on it.
A practical split for a typical hypertrophy block:
| Exercise type | Reps | Prescribe | Collect the rating? |
|---|---|---|---|
| Main compound | 3–6 | 2 RIR, dropping to 1 late in the block | Yes, every working set |
| Secondary compound | 6–10 | 2 RIR | Yes, last set only |
| Accessory | 10–15 | 1–2 RIR | Last set only |
| Isolation | 15–20 | 0–1 RIR | No, just log the reps |
The point of the last column is that a rating you do not read is admin you inflicted on a client for nothing. Ask where you will act on the answer.
A worked block
Here is the same four-week accumulation block written in RIR, for a client eighteen months in.
| Week | Main compound | Accessories | What you are watching |
|---|---|---|---|
| 1 | 4 × 6 at 3 RIR | 3 × 12 at 2 RIR | Are the reported numbers plausible against the load? |
| 2 | 4 × 6 at 2 RIR | 3 × 12 at 2 RIR | Load should rise slightly at the same reported effort |
| 3 | 4 × 6 at 1 RIR | 3 × 12 at 1 RIR | Load flat and effort climbing means fatigue is ahead of plan |
| 4 | 4 × 6 at 1 RIR | 3 × 12 at 0–1 RIR | Reported effort at week-two loads is the deload signal |
Volume and exercise selection do not move. Only the reserve does, one rep per week, and the client's reported numbers tell you whether the block landed. If effort is climbing while the weights sit still, the deload is due before the calendar says it is — how to program deload weeks covers what to cut and by how much.
Written in RPE the same block reads 7, 8, 9, 9–10. Identical prescription, harder question. That is the entire trade, and for most clients it is not worth making. The longer method for teaching a client to rate accurately, including what to do when their numbers are obviously wrong, is in how to use RPE with online clients.
