Training to Failure — How Close Should You Go?

Last set of bench press.

The rep slows down. You are fairly sure one more will go — about the one after that, you are no longer sure.

Do you rack it now or keep pressing?

The old gym answer was simple: if you do not go to failure, you are leaving muscle growth on the table.

The current research paints a more interesting picture.

Proximity to failure matters for hypertrophy. Absolute failure is still not necessary in every set.

That difference sounds small, but it changes training practice considerably.

What “muscular failure” actually means

A clean definition matters for this article.

Momentary muscular failure means: you attempt another rep with maximum effort but can no longer complete it through the intended movement.

Two things should be distinguished from that:

Technical failure: The rep might somehow still be possible, but no longer with the intended technique.

Voluntary set end: You stop even though several clean reps were still available.

In practice, many lifters steer their sets by reps in reserve (RIR).

If you stop after ten reps and realistically could have got two more clean ones, the set was at roughly 2 RIR.

→ RPE table

→ RIR vs. RPE: what’s the difference?

What the research shows

The most important statement first:

Training to absolute muscular failure is not a prerequisite for strength or muscle growth.

Several meta-analyses and the 2026 ACSM position stand confirm that.

At the same time, the reverse claim — “1–3 RIR is always exactly as good as failure” — would be too strong as well.

The meta-regression by Robinson et al. (2024) found a relationship for hypertrophy in the direction of more muscle growth the closer sets ended to failure. For strength, that relationship was negligible.

That fits the earlier meta-analyses by Refalo et al. and Grgic et al.: failure is not strictly necessary, and the difference between failure and non-failure comes out small overall. How close the non-failure sets actually were to failure is decisive here.

In practice that means:

• 5–6 RIR and 1–2 RIR are not simply the same training.
• 1 RIR and 0 RIR, by contrast, probably differ less in hypertrophy stimulus.
• The last possible rep can add stimulus, but it also costs additional fatigue.

So the sensible range is not a single magic RIR value.

Why proximity to failure matters at all

With heavy loads, many motor units have to be recruited from the start, because the force requirement is high.

With lighter loads, a set begins at a lower relative demand. As fatigue builds, additional motor units have to contribute so you can keep performing reps.

Which is why higher rep ranges work for muscle growth too — when the set progresses far enough.

That is where the popular idea of the “last five effective reps” came from.

As an illustrative model it is usable. As a hard biological boundary it is too simple.

There is no switch that flips exactly five reps before failure, after which every rep produces the same amount of growth.

This picture is better:

As you get closer to failure, the relevant muscle fibre demand rises in many sets. But the additional benefit does not grow without limit, while fatigue and technique problems can keep increasing.

Which is exactly why “hard enough” is more useful than “always until nothing is left”.

Failure has a price

A set to failure is not automatically harmful.

It is simply often more fatiguing than a comparable set that ends slightly earlier.

Studies on acute recovery show larger performance drops after failure protocols, and slower recovery of certain neuromuscular and metabolic markers, than after sets ended submaximally.

That is particularly relevant if you still have several exercises to go afterwards, or want to train the same muscle group again within a few days.

A single additional rep can look sensible in isolation.

But the more important question is:

What happens to set 2, set 3, the next exercise and the next session?

Training does not consist of one single maximal set.

Where failure can make sense — and where a buffer is usually more practical

The research prescribes no universal table under which curls must end at exactly 0 RIR and squats at exactly 3 RIR.

Still, a sensible practical heuristic can be derived from risk, fatigue and the character of the exercise.

Isolation work and stable machines

Biceps curls, leg extensions, lateral raises or a well-adjusted chest press have one advantage:

A failed rep is usually easy to control.

So here you can go very close to failure more often — say 0–2 RIR — and occasionally actually train to momentary failure.

That does not mean every set has to fail.

It only means the cost of a failed rep is often lower.

Heavy free-weight compound lifts

On squats, deadlifts and heavy bench press, the arithmetic changes.

Under high fatigue, technique, stability and safety can become the greater limiters here. At the same time, heavy compound lifts generate a lot of systemic load.

A buffer of roughly 1–3 RIR, or more on very heavy work, is therefore sensible for many sets.

Not because muscle suddenly stops growing at 0 RIR.

But because you are weighing the additional stimulus against the cost to the whole session.

Very heavy strength work

At low rep counts, absolute proximity to failure is not the only purpose of the exercise for muscle growth anyway.

When you train heavy doubles or triples, technique under load, force production and specific practice of the movement can be the priority.

There, a clean submaximal set is often more valuable than a forced grind.

How to estimate your RIR better

RIR is not a perfectly objective measurement.

But the research shows something practical: estimates get more accurate closer to failure.

In the study by Remmert et al. (2023), predictive accuracy depended more on proximity to failure and position within the set sequence than on sex or training experience.

That contradicts the often-repeated blanket claim that beginners are generally off by “3–5 reps” while experienced lifters are automatically precise.

RIR can be learned, but not felt perfectly.

Three methods help.

1. Calibrate occasionally

Pick a safe exercise — chest press, leg extension or cable curl, for instance — and occasionally take a set to actual momentary failure.

If you thought beforehand you were at 2 RIR and then get five more reps, you have gained a useful reference.

That is calibration, not a recommendation to structure every session that way.

2. Watch movement speed

The closer you get to failure, the more concentric speed typically drops at maximum intent.

A slow rep does not automatically mean “1 RIR” — different exercises and people have different velocity profiles.

But a clear loss of speed is a good signal that the set is entering a very demanding range.

3. Treat RIR as a range

You do not have to decide whether a set ended at mathematically exactly 1.0 or 1.8 RIR.

For practical training, this is often enough:

• far from failure
• moderately demanding
• close to failure
• actual failure

That stops a useful tool from turning into false precision.

The short version

Training to failure is a tool — not a mark of quality.

For hypertrophy:

• Proximity to failure counts. Very easy sets do not automatically deliver the same stimulus as hard sets at the same volume.
• Absolute failure is not necessary. The additional benefit over training ended very close to failure is probably small and situation-dependent.
• 1–3 RIR is a sensible default for many working sets, but not a biologically exact optimal zone.
• On stable isolation and machine exercises, 0 RIR can be defensible more often.
• On heavy free-weight compound lifts, an additional buffer is often the better cost-benefit decision.
• RIR estimates tend to be more precise closer to failure. Occasional safe failure sets can help with calibration.

How this set-level effort interacts with training volume matters at least as much:

→ What are effective sets?

→ Hypertrophy training: guide for muscle growth

So the practical target is not:

“Max out every set.”

It is:

“Train close enough to the limit that the set is productive — and stay far enough away that the rest of your training still works.”

Sources

• Currier BS, D’Souza AC, Fiatarone Singh MA et al. (2026). Resistance Training Prescription for Muscle Function, Hypertrophy, and Physical Performance in Healthy Adults: An Overview of Reviews. Medicine & Science in Sports & Exercise, 58(4):851–872.
• Robinson ZP, Pelland JC, Remmert JF et al. (2024). Exploring the Dose–Response Relationship Between Estimated Resistance Training Proximity to Failure, Strength Gain, and Muscle Hypertrophy: A Series of Meta-Regressions. Sports Medicine, 54(9):2209–2231.
• Refalo MC, Helms ER, Trexler ET et al. (2023). Influence of Resistance Training Proximity-to-Failure on Skeletal Muscle Hypertrophy: A Systematic Review with Meta-analysis. Sports Medicine, 53(3):649–665.
• Grgic J et al. (2022). Effects of resistance training performed to repetition failure or non-failure on muscular strength and hypertrophy: A systematic review and meta-analysis. Journal of Sport and Health Science, 11(2):202–211.
• Morán-Navarro R, Pérez CE, Mora-Rodríguez R et al. (2017). Time course of recovery following resistance training leading or not to failure. European Journal of Applied Physiology, 117(12):2387–2399.
• Remmert JF, Laurson KR, Zourdos MC (2023). Accuracy of Predicted Intraset Repetitions in Reserve (RIR) in Single- and Multi-Joint Resistance Exercises Among Trained and Untrained Men and Women. Perceptual and Motor Skills, 130(4):1682–1698.