How many reps in reserve for hypertrophy should you keep?

The usual mistake is treating every set as a test. Lifters grind each set of ten to the last possible rep, believing growth lives there, and pay for it on every set that follows. For hypertrophy, keep most working sets one to three reps in reserve (RIR), meaning one to three more clean reps were left when you racked the bar [1]. Take a set to zero only on the last set for a muscle, and mostly on simple isolation lifts [1]. Four reps in reserve is the furthest a set can sit from failure and still count as effective work in the app.

What the failure studies actually show

Training to failure does not beat stopping short of it for muscle growth. A meta-analysis of 15 studies found no advantage for sets taken to momentary muscular failure over sets stopped before it [2]. Across every definition of failure the studies used, the edge for failure was trivial, an effect size of 0.19 [2].

The cleanest test came next. In an eight-week trial, 18 trained lifters trained one leg to failure and the other to a perceived 2 RIR on the leg press and 1 RIR on the leg extension [3]. Quadriceps thickness rose 0.181 cm on the failure side and 0.182 cm on the RIR side, and the failure side lost more reps and bar speed within each session [3]. Same growth, more fatigue.

That does not mean distance from failure is free. A series of meta-regressions that estimated RIR for every study found muscle size grew more as sets ended closer to failure, while strength gains stayed similar across a wide range of RIR [4]. The authors call the analysis exploratory and say the exact curve is still unclear [4]. Read together, the studies point to the same zone: close enough that the last reps are hard, not so close that the next set suffers.

Where a set stops counting

The app scores each logged set for effective reps, and the cutoff is sharp. A set of ten at RPE 6, which is 4 RIR, counts all ten reps. The same ten reps at RPE 5.5, which is 4.5 RIR, count zero. A set logged with no RPE also counts zero effective reps, because the app cannot tell how hard it was. Warm-ups never count, whatever their rating.

Four reps in reserve is a floor, not a target. It is the line below which a set is too easy to call work. The useful range sits above it.

Lifters also misjudge the number. In a meta-analysis of 414 participants, people underpredicted how many reps they had left by 0.95 reps on average, and their guesses got better closer to failure and in later sets [5]. A set you call 2 RIR may well have been 3. That is one more reason not to aim for 4: a set rated 4 might really be 5, and that is junk volume.

A worked example at 10 reps

Take a lifter whose estimated 1RM on the bench press is 100 kg, working in sets of ten. The model behind the one-rep max calculator gives the load for each effort level.

Those are exact outputs before rounding to the plates you own. The whole effective range, from 4 RIR to failure, is a narrow band of load. Walking from the 3 RIR row to the 1 RIR row is what a block ramp does to the bar over a few weeks.

Logging the rating honestly matters for the next weight too. A set of 75 kg for ten logged at RPE 8 estimates a 110.3 kg max. Log the same set at RPE 9 and the estimate is 106.1 kg. Over-rating effort makes a lifter look weaker than they are, and the next weight lags.

How a hypertrophy block ramps it

In a program, the Hypertrophy block starts its first week further from failure and lowers the RIR target week by week, then ends on a deload. Early weeks leave room to find loads and add sets. Later weeks spend that room. The RIR progression scheme then steers the load by the effort you log against each week's target, and a deload week never emits an increase.

Weekly volume and effort trade off. Pushing every set to failure tends to cost reps on the sets after it [3], so the same session holds fewer good sets. For how many sets to aim at, see how many sets per week per muscle.

Questions

Is 0 RIR better than 2 RIR for hypertrophy?

Not in the trials so far. Trained lifters who stopped at 1 to 2 RIR grew their quadriceps as much as when they went to failure, and they lost fewer reps and less bar speed in each session [3]. Failure has a place on the last set of an isolation lift, where the fatigue costs little. On heavy compound lifts it mostly makes the following sets worse.

Does a set at 5 RIR build any muscle?

It may build some, but the app does not count it. A set at RPE 5, five reps short, scores zero effective reps, and so does any set logged without a rating. The research also leans the same way: muscle size grew more as sets ended closer to failure [4]. If a set felt that easy, add load next time rather than more sets.

How do I know if my RIR guesses are right?

Test them now and then. Rate a set, then keep going to failure with a spotter or on a safe machine and count the reps you actually had. Lifters underpredict by about one rep on average, and accuracy improves closer to failure and in later sets [5]. Repeat the check every few weeks, and your ratings, and your logged effort, become far more useful.

Should RIR differ between compound and isolation lifts?

Yes. Keep squats, presses and deadlift variations a few reps short most of the time, and save 0 RIR for the final set of simple, low-risk lifts such as curls or leg extensions [1]. A missed rep on a lateral raise costs nothing. A grinder on a heavy squat costs the back-off sets and some recovery for the next session.