RPE for beginners: how to rate a set and log it honestly

The usual beginner mistake with RPE is rating a set as harder than it was. When lifters in a pooled set of 12 studies predicted how many reps they had left and then kept going to failure, they came up short by 0.95 reps on average [1]. RPE for beginners comes down to one rule: the number counts the reps you had left when you racked the bar. RPE 10 is none, 9 is one, 8 is two.

The scale takes a minute to learn. Rating a set honestly takes a few weeks of practice, and that practice can be planned like any other part of a program. Below is what each number means in the app, what happens to your estimated 1RM when a set goes in with no rating at all, and a two-week drill that shows you how far off your own guesses are.

RPE counts the reps you had left

RPE stands for rating of perceived exertion. The version lifters use today ties each number to reps in reserve (RIR): how many more reps you could have done with good form at the moment you stopped. On that scale RPE 10 means no reps in reserve, RPE 9 means one, and each step down adds a rep [2]. The reps-left descriptions run from RPE 10 down to RPE 5. Below that, the scale stops counting reps and describes how light the set felt [3].

The app uses the same scale and prints the reps-left meaning beside each value. The effort control you tap after a set has four stops, and each one writes a fixed RPE. You can also type a value, and the app keeps it in half steps between 1 and 10.

The half steps exist for sets you are honestly unsure about. RPE 9.5 says you might have squeezed out one more rep, or you might have missed it. Use it for that case only. A beginner who logs 9.5 on every hard set is recording a shrug, and the log cannot tell the difference.

Two definitions matter before you rate anything. Failure means the rep you could not finish without your form breaking, which is how the research defines it too [4]. A rep that needs a bounce off the chest or only gets halfway is a failed rep. And the rating belongs to the set you just did, at the moment you racked it, not to how the session felt overall.

A set without a rating counts as a max

Leave the effort blank and the app treats the set as RPE 10, as if you could not have done another rep. Every classic one-rep max formula makes the same assumption, because a formula that sees only weight and reps cannot know what you left in the tank. That makes a missing rating the most conservative thing you can log.

Here is what the rating does to one set. Take a lifter who benches 50 kg for 8 reps. The estimated 1RM below comes from the app's model, the reps by RPE chart read along its diagonal, and the same numbers come out of the one-rep max calculator if you enter the set yourself.

Same bar, same reps, five different lifters on paper. The rating is the only thing that changed.

The estimate then sets loads. Suppose the honest rating was RPE 8. The model puts this lifter's 8-rep max at 53.2 kg, and a fresh set of 8 at RPE 8 at 50 kg, which is the set they just did. That checks out.

Now suppose the same set is logged as Easy. The estimate climbs to 73.5 kg, and the model's answer for 8 reps at RPE 8 becomes 54.3 kg. That load is heavier than the 8-rep max the honest rating implied. The lifter walks up to a set of 8 that, by their own body's numbers, ends in a missed rep before 8.

Logged as Fail or left blank, the estimate drops to 63.6 kg and the same target comes out at 47 kg. That is a lighter session than the lifter could have done, and nothing worse. Rating a set too hard costs you a little progress. Rating it too easy costs you reps you cannot complete.

So when you really do not know, Fail or blank is the safe answer. When you do know, say so, because a log full of blanks will always plan a little light.

Which way beginners get it wrong

The research points in one direction: lifters think they have fewer reps left than they do. In one study, 141 gym members predicted their reps to failure on six familiar exercises, five of them on machines, and then did the set to failure. On average they underpredicted, and the standard error of their guesses ran from 2.64 to 3.38 reps depending on the exercise [4]. Accuracy tended to improve with training experience in that study.

The pooled data is less tidy. Across 12 studies and 414 participants, the average miss was 0.95 reps on the short side, and training status did not clearly change it [1]. What did help was proximity: predictions made closer to failure were more accurate, and so were predictions on sets of 12 reps or fewer [1]. Experience may matter less than how close you are to the end of the set when you make the call.

Novices also misread the top of the scale. In the study that introduced the reps-left scale, experienced squatters rated their true 1RM at an average of 9.80, while novices rated the same all-out single at 8.96 [3]. A novice's maximum felt to them like it had a rep left in it. The experienced group also moved their maxes more slowly, which the authors read as a sign they could grind through a heavy rep that a novice would stop short of [2].

Put together, a beginner's ratings are most likely to be too high in effort, least reliable on long sets far from failure, and least reliable of all on a max. That is why the authors who laid out how to use the scale suggest that novices record ratings alongside their normal training without basing load decisions on them alone until their accuracy is established [3]. You can find out where you stand in two weeks.

A two-week drill to calibrate your ratings

The test is the one the scale's authors suggest: rate a set before it ends, then take the same load to failure and see whether the rating held [3]. Run it on purpose for two weeks and you get a record of your own error.

Pick two exercises where a failed rep is safe and boring. A machine chest press, a leg press, a cable row and a dumbbell curl all work, and they resemble the exercises in the study above [4]. Skip barbell squats and bench presses unless you have safety arms set at the right height. Use a load you already train with, one that lands somewhere between 6 and 12 reps, since that is where predictions are most accurate [1].

On the last set of each of those exercises, every session, do this. Train the set as normal. At the rep where you would usually stop, make your call silently: how many more reps you think you have. Then keep going, with clean form, until you cannot finish a rep. The reps you added are your real reps in reserve. Write both numbers down, the guess and the result.

Log the set in the app as what it became: the full rep count at RPE 10. The drill set was taken to failure, and the log should say so. Keep the guess-against-result record somewhere else, on paper or in a note on your phone.

After two weeks, read the record. If your guesses mostly land within one rep of the result, your ratings are good enough to steer load, and you can rate working sets with confidence. If you are regularly two or more reps short, keep rating every set, but let the rep count and the load drive your progression for another few weeks, then run the drill again.

A practical note on the other sets in that session: do not take everything to failure because the drill is running. One drill set per exercise is the whole cost. The rest of the session stays as written.

Habits that keep the rating honest

Rate the set the moment you rack the bar. By the end of the rest period the memory of that last rep has already softened, and a set that ground to a halt starts to feel like it had one more in it.

Rate working sets, not warm-ups. Warm-ups never count toward volume in the app and their effort tells you nothing about the next weight, so leave them blank.

Do not use a rating to rescue a missed rep. If you aimed for 8 and failed on the eighth, the set is 7 reps at RPE 10. Logging 8 reps at RPE 9.5 flatters the estimate and plans the next session too heavy.

Treat a form break as the end of the set. If the last rep needed your hips off the bench, the rep before it was your RPE 10. Rating by the ugly rep teaches you to count reps the next set will not be able to repeat.

Be careful with high-rep sets. Once a set runs past 12 reps, predictions get worse [1]. A set of 20 rated as RPE 8 is a guess with a wide error, so give it less weight than a set of 6 at the same rating.

Pick a target rating before the set, not after. If the plan says RPE 8, you are looking for two reps left, and the moment you feel them is the moment to rack the bar. For how hard a top set should be across a block, the top set RPE answer covers the usual targets, and estimated 1RM explained shows the full curve behind the numbers in the table above.

How ratings change your progression

Of the seven progression schemes in the app, Double progression decides mainly on reps: hit the top of the rep range on every set and the load goes up. It works on day one, before your ratings mean much, which makes it a sensible default while you learn the scale. The RIR scheme steers load by the effort you log against a target, so it is only as good as your ratings. Move to it after the drill says your guesses hold up.

Whichever scheme you run, the honest log is what lets it work. A lifter who rates every working set, even roughly, gives the app a better picture than one who rates the good days and leaves the bad ones blank.

Questions

Should a beginner use RPE at all?

Yes, as a habit, with limits. Rate every working set so you learn the scale, but let reps and load decide your progression until a calibration drill shows your guesses land within about a rep of the truth. The authors who set out how to use the reps-left scale recommend exactly that order for novices: record the ratings first, and rely on them for load only once accuracy is established.

What does RPE 8 mean?

RPE 8 means you finished the set with two good reps left. You could have done two more with the same form, and a third would most likely have failed. It is the Good stop on the app's effort control and the most common target for working sets. If the last rep slowed down noticeably but you were confident of two more, RPE 8 is the right call.

Is it bad to leave RPE blank?

No. A blank rating is treated as RPE 10, so the estimated 1RM from that set is the lowest the set can produce. Your next weights will come out a little light, never too heavy. The cost of logging everything blank is slower progress, because the app cannot see the sets where you had reps to spare. Rate when you know and leave it blank when you honestly do not.

How long does it take to learn RPE?

There is no fixed number. In the research, accuracy tended to improve with training experience in some studies and barely changed with it in pooled data, while predictions made close to failure were reliably better. A two-week drill on safe machine exercises gives you your own answer: if your guesses land within a rep, you have learned it well enough to use. If not, run the drill again in a few weeks.