How Close to Failure Should You Train to Build Muscle?

TL;DR: For hypertrophy, take most of your working sets to within 0–3 reps of failure. Going to true muscular failure produces only a sliver more growth than stopping a couple of reps short — and it costs disproportionately more fatigue. Stopping 4 or more reps shy of failure, on the other hand, leaves real gains on the table. Train hard, estimate your reps in reserve honestly, and treat proximity to failure as a dial you adjust by lift and by phase.

---

"How hard should each set be?" is the question that decides whether your volume actually produces muscle. You can do all the right sets at the right frequency, but if you stop them too early, you under-stimulate growth — and if you grind every one into the ground, you bury yourself in fatigue you can't recover from. The good news: the research on proximity to failure is now clear enough to give you a working answer. Here it is, and here's how to apply it.

What "Reps in Reserve" Actually Means

Reps in reserve (RIR) is simply the number of reps you could still have done at the end of a set before hitting failure. Stop a set when you're sure you had two more clean reps in you, and that's 2 RIR. Grind until the bar stops moving and you physically cannot complete another rep — that's 0 RIR, or true momentary muscular failure.

RIR is the practical, autoregulated version of effort. It's the inverse of the RPE (rate of perceived exertion) scale that Helms and colleagues adapted for resistance training: an RPE of 8 means roughly 2 reps in reserve, a 9 means 1, a 10 means you went to failure (Helms et al., 2016). Whichever language you use, the point is the same — you're rating how close each set came to the wall, then using that rating to steer your training.

The Research: Closer to Failure Helps, but With Diminishing Returns

For years the debate was binary: do you have to train to failure to grow, or not? The current evidence reframes it as a dose-response curve, and the shape of that curve is what matters.

A systematic review and meta-analysis by Refalo et al. (2023) pooled the controlled studies comparing different proximities to failure and found a small advantage for training closer to failure for hypertrophy — effect sizes on the order of 0.15 to 0.21. Real, measurable, but modest. Robinson et al. (2024) went further with a series of meta-regressions modeling the relationship directly, and found that estimated muscle growth does increase as sets are taken nearer to failure — while strength gains are far less sensitive to how close you stop. In other words, proximity to failure is mostly a hypertrophy lever, not a strength one.

Then there's the practical test. Refalo et al. (2024) randomized 26 resistance-trained adults to either train every set to momentary failure or stop with 1–2 reps in reserve for eight weeks. Quadriceps hypertrophy was essentially the same between groups. Training to absolute failure bought nothing extra over leaving a rep or two in the tank — but it demanded more from recovery to get there.

Put the three findings together and the picture is consistent: growth keeps climbing as you approach failure, but the curve flattens hard in the last couple of reps. The difference between 0 and 2 RIR is trivial for muscle. The difference between 2 and 6 RIR is not.

The Practical Window: 0–3 RIR

For most working sets aimed at building muscle, train to 0–3 reps in reserve. That single range captures nearly all the available stimulus while keeping fatigue manageable. Where you land inside it isn't random — it should track the exercise and the phase you're in:

• Heavy compounds (squat, deadlift, big presses): keep 1–3 RIR. The injury risk and systemic fatigue of grinding a near-maximal barbell lift to failure aren't worth the marginal growth, and failure wrecks your bar speed and technique on the reps that follow.
• Machines and single-joint isolation (leg press, curls, lateral raises, extensions): here you can push to 0–1 RIR safely. There's no technical breakdown or pin to escape, so failure is cheap to reach and easy to recover from. This is where occasional true failure earns its place.
• Early in a training block: sit at the higher end (2–3 RIR) so you have room to ramp intensity as the weeks progress. Late in a block, before a deload: drift toward 0–1 RIR to peak the stimulus, then back off.

The one rule that holds across all of it: a set 4+ reps shy of failure is mostly a warm-up. If you're routinely finishing sets you could have extended by five reps, your problem isn't your program — it's your effort.

Why You're Probably Further From Failure Than You Think

Here's the catch that undermines most lifters' RIR: people are bad at it, and they err in one direction. When you feel like you have two reps left, you often have four or five. Untrained and even intermediate lifters consistently stop short of where they think they are, which means a program written for "2 RIR" is frequently being executed at 5+ RIR — squarely in warm-up territory.

Two fixes. First, calibrate occasionally: on an isolation movement, actually take a set to true failure and count how many reps came after the point you'd normally have racked it. That recalibrates your internal gauge. Second, log it. Estimating RIR from memory at the end of a workout is guesswork; recording it set by set turns it into data you can adjust against — and accuracy improves quickly once you're paying attention.

Make Proximity to Failure a Dial, Not a Default

The lifters who progress fastest don't pick one effort level and hold it forever. They run heavy compounds at 1–3 RIR, push isolation work to 0–1, start blocks conservative and finish them hard, and they pair that with the other levers that actually drive growth — enough weekly volume, real progressive overload over time, and adequate rest between sets so each set is genuinely hard rather than cardio. If you haven't structured those pieces yet, start with how to build a workout program from scratch and layer RIR on top. (Choosing loads that land you in the right rep range near failure is easier when you know your maxes — the one-rep max calculator gets you a working estimate.)

This is exactly the kind of thing that's hard to hold in your head and easy to drift on. Protokl logs the RIR on every set, watches how your performance trends across a block, and adjusts your prescribed loads and effort so you stay in the productive window instead of guessing — turning "how hard should this set be?" from a daily judgment call into something the program quietly manages for you. Train hard on purpose, not by accident.