How close to failure should you train for hypertrophy?

For muscle growth, end most sets 1 to 3 reps short of failure. Leave about 2 to 3 reps in reserve on heavy compound lifts such as squats and deadlifts, and 0 to 1 on isolation and machine lifts, where going to failure is safe and cheap to recover from. Save true failure for the last set of a safe exercise.

Our guide on how many sets and reps to build muscle counts hard sets. A hard set is a working set that ends within a few reps of failure. That leaves one question open: how few is a few? This post answers it for each type of lift, for your goal and for your experience.

What does 'close to failure' actually mean?

Failure is the point where you cannot do another clean rep. Reps in reserve (RIR) means how many more clean reps you could have done when the set ended.

RPE is the other way to say it. RPE stands for rate of perceived exertion, how hard the set felt on a scale of 1 to 10. RPE 8 means about 2 reps left in reserve. RPE 10 means nothing left.

Picture the 8th rep of a squat. It slows down, but you know 2 more would come. Rack it there, and that set ended at about 2 reps in reserve, or RPE 8.

Our post RPE explained covers how to rate and log it. Here we only care about where to stop.

Does training closer to failure build more muscle?

A little, probably. The evidence points that way, but it is soft.

Refalo and colleagues (Sports Medicine, 2023) pooled the studies in a meta-analysis. They found no significant difference in muscle growth between sets taken to failure and sets stopped short of it. That fits earlier work: 2 systematic reviews from 2021 also found no hypertrophy benefit from training to failure.

Robinson and colleagues (Sports Medicine, 2024) asked a slightly different question. Their meta-regressions showed hypertrophy improves as sets end closer to failure. Strength gains were similar across a wide range of reps in reserve.

Both papers say the exact relationship stays unclear. Most studies estimated reps in reserve rather than measuring it. So treat the evidence as low certainty.

Putting it together: the closer you get, the more growth you are likely to get. The extra from the last 1 or 2 reps is small.

Is stopping 1 to 2 reps short as good as going to failure?

For the quadriceps, one good study says yes.

In Refalo and colleagues (Journal of Sports Sciences, 2024), trained men and women ran 8 weeks of quadriceps training. Sets ended at 1 to 2 reps in reserve grew the quadriceps about as much as sets taken to failure.

Reaching failure caused bigger drops in lifting speed. That is fatigue. The extra fatigue did not pay off in extra muscle.

It is one study on one muscle. It does not settle every muscle group. It does make 1 to 2 reps short a safe default for most of your sets.

How many reps in reserve should you leave on compound and isolation lifts?

Not the same number on every lift. The heavier and riskier the lift, the more you leave.

The first reason is safety. When a squat or a bench fails, the bar has to land somewhere. Our bench press guide is plain about it: failure without a spotter or safety arms is a safety problem. When a curl fails, your arms stop and you lower the dumbbells.

The second reason is recovery. A heavy set of squats taxes your legs, your back and your grip at once. A set to failure there can blunt the next 2 sets and the next session.

The third reason is form. On a deadlift, your back tends to round before your glutes and hamstrings are done. The reps near failure train the habit you are trying to avoid. On a leg curl, the machine holds your position, so the last rep is still a clean rep.

Does it change with your goal and experience?

Yes, a little.

Muscle first: use the table above.

Strength first: Robinson's review found strength gains were similar across a wide range of reps in reserve. Leave 2 to 3 on the main lifts. You keep the bar speed and the practice, and you skip the grinding.

New to lifting: you will misjudge reps in reserve. Everyone does at first. In one study, trained men underestimated it by about 2.3 reps on the bench press and 1.4 on the prone row. They thought they were closer to failure than they were. Leave 2 to 3 and test yourself now and then.

Where you are in a plan matters too. Coaches work in a training block, a plan of 4 to 12 weeks. Early in the block, start further from failure, around 3 reps in reserve. Late in the block, get closer, around 1. The weight climbs while the effort catches up.

When is going all the way to failure worth it?

In 2 cases, and both are on safe lifts.

The last set of an isolation or machine exercise. Take the final set of lateral raises or leg curls to failure. You get a little extra stimulus, and you check your guesses at the same time.

A rep-out test every few weeks. Pick a safe exercise, such as a machine row. Do a set you would call RPE 8, then keep going until you cannot do another clean rep. That teaches you what RPE 8 really feels like.

The test has a trigger and a result. If a set you called 2 in reserve turns into 5 more reps, you have been stopping too early. Add weight or reps next session.

When failure is not worth it:

• Heavy barbell compounds: the risk and the recovery cost outweigh a small gain.
• Every set of the session: fatigue piles up and later sets get worse.
• The week before a deload: a deload is a lighter week that lets fatigue drop. Our post on the deload week covers when to take one. Do not dig the hole deeper right before it.

How do you know you are close enough?

Watch rep speed. Near failure, the last rep slows down without your form changing. If your form changes first, you went too far for that lift.

Here is a worked example. Lateral raise, target 12 to 15 reps at 1 rep in reserve. You get 15 reps and could have done 1 more. Log RPE 9.

Then use double progression. Our progressive overload guide covers it in full. Stay at the same weight inside the rep range. Once every set hits 15 with clean form, add the smallest weight step next session. Then work back up from 12.

Do not react to every set. One rough set is not a signal. A pattern across a week is. If your sets at RPE 9 keep landing well short of the range, the weight is too heavy. If they keep landing past it, the weight is too light.

How Trackist runs it

You log weight, reps and RPE (1 to 10) for each set. From your second session of an exercise, a line above it suggests today's weight. Trackist works it out with fixed rules from your last session's sets, target reps and logged RPE.

Hit the top of the rep range on every set, and Trackist adds a step that fits the equipment: 2.5 kg on a barbell, 2 kg on dumbbells, 5 kg on a machine or cable. Fall short, and it keeps the weight, takes about 5% off or leaves the call to you. That depends on how far short you fell and how hard the sets felt. If you miss reps at RPE 10, it keeps the weight rather than cutting it.

Trackist does not change the weight between sets. The suggested weight stays the same for the whole session and moves from one session to the next. You choose and log what you actually lift.

If a session was too hard, say so. The first report is noted. The same report in a later week rewrites the day or rebuilds the rest of the plan. Our post on a plan that is too hard or too easy explains what to change yourself.

An honest limit: the app has no RIR targets. You use the table above and rate each set honestly. An AI trainer is only as good as what you log. Guessed numbers or the same RPE on every set give the suggestions nothing real to work from.

Want a plan to try this on? Our 8-week hypertrophy plan runs 4 days a week.

Log your next session in Trackist. Trackist writes a plan around your week and your gym, and suggests your next weight from the sets you logged.