Do you need to train to failure to build muscle and strength? What the research shows, why stopping 1 to 3 reps short works, and when failure makes sense.
Most of your sets do not need to go to failure. Research suggests stopping 1 to 3 reps short (1 to 3 reps in reserve) builds muscle about as well as going all the way, with less fatigue and lower injury risk on big lifts. Save true failure for the last set of safer exercises like curls, machines, and cable work, and keep heavy squats, deadlifts, and bench at 1 to 2 reps in reserve.
Few training questions start more arguments than this one. One camp says if you did not fail, you did not grow. The other says failure is a recipe for burnout and injury. Both have a point, and neither has the full picture.
What the research supports is more useful than either extreme: sets need to be hard, but they rarely need to be all the way to failure. The gap between "close" and "failure" matters less than most people think, and the fatigue it costs matters more.
Failure usually means momentary muscular failure: the point where you cannot complete another full rep with good form, despite trying. It is not the point where the set gets uncomfortable, and it is not the point where you decide to stop.
That distinction matters, because many lifters who think they train to failure are actually stopping 2 to 4 reps short. A useful way to describe how close you came is reps in reserve (RIR): how many more clean reps you could have done.
| RIR | What it means | How it feels |
|---|---|---|
| 0 | Failure, no more reps possible | Max effort, bar slows to a crawl |
| 1 | One more rep was there | Hard, last rep was slow |
| 2 | Two more reps were there | Challenging but controlled |
| 3 | Three more reps were there | Moderate, speed still decent |
| 4+ | Well short of failure | Easy to very easy |
Work by Zourdos and colleagues in 2016 showed that an RIR-based scale tracks bar speed reasonably well, and that experienced lifters estimate it more accurately than novices. Helms and colleagues described how to apply this scale in program design the same year.
The best current answer comes from a 2023 meta-analysis by Refalo and colleagues, which pooled studies comparing sets taken to failure against sets stopped short. They did not find a clear hypertrophy advantage for training to failure. Stopping a little short produced similar growth.
A related finding helps explain why. A 2021 network meta-analysis by Lopez and colleagues found that low, moderate, and high loads produced similar muscle growth when sets were taken to or near failure. In other words, effort is a key ingredient for growth, but "near failure" seems to cover it. You do not need to squeeze out the last grinding rep for a set to count.
These studies vary in how they define and measure proximity to failure, and lifters are not perfect at estimating RIR. The honest summary is that a range of roughly 0 to 3 RIR appears to work well for growth. Sets stopped with 5 or more reps left in the tank are probably less effective, though they still count for something.
Strength depends heavily on practicing heavy, well-executed reps. Grinding to failure on a heavy single or triple adds a lot of fatigue and tends to degrade technique on exactly the reps you want to be practicing.
Lopez and colleagues also found that heavier loads were better for maximal strength. Combine that with the fatigue cost of failure, and the practical answer for strength work is to lift heavy, but leave a rep or two in the tank on most sets.
Failure is not dangerous on every exercise, but it is expensive.
Failure is a tool, not a sin. Use it where the risk is low and the information is useful.
A simple approach that works for most people:
Here is what that looks like over four sessions of dumbbell bench with a 6 to 10 rep range:
| Session | Weight | Reps (3 sets) | Last-set RIR | Next step |
|---|---|---|---|---|
| 1 | 60 lb | 8, 8, 7 | 1 | Add a rep |
| 2 | 60 lb | 9, 9, 8 | 1 | Add a rep |
| 3 | 60 lb | 10, 10, 10 | 2 | Top of range, add weight |
| 4 | 65 lb | 7, 7, 6 | 1 | Build reps again |
No set went to failure, and the lifter still added weight within three sessions.
Alexandros is an iPhone app that acts as a personal weightlifting coach you talk to in a chat. Every set you log gets a five-point effort rating: Very easy, Easy, Moderate, Hard, or Max effort. Those map to reps in reserve (roughly 4+, 3, 2, 1, and 0), or you can enter RIR directly.
That rating drives the recommendation. If you hit the top of your rep range with at least one rep in reserve, it suggests adding weight. If you are within the range, it suggests adding a rep, or two if the set felt very easy. If a set was a grind and fell below the range, it backs the load off by about 8%. You get progression without needing to hit failure, and it explains the why if you ask it to.
No. Research suggests that stopping 1 to 3 reps short of failure builds muscle about as well as going to failure, with less fatigue. The sets do need to be genuinely hard, though.
It is not automatically harmful, but it is usually inefficient. Fatigue builds quickly, later sets suffer, and heavy compound lifts become riskier. Most lifters do better reserving failure for the last set of safer exercises.
Practice. Pay attention to bar speed, since reps slow sharply as you approach failure. Occasionally take a safe exercise to failure and compare it to your prediction. Experienced lifters tend to estimate more accurately than beginners.
Beginners are less accurate at judging how close they are to failure and are still learning technique. Keeping most sets at 2 to 3 reps in reserve is a safer way to build both skill and muscle.
