Proximity to failure: RIR and intensity for hypertrophy

An evidence-grade summary of how close working sets need to be taken to muscular failure to drive hypertrophy, the role of reps in reserve as a load-prescription tool, and the trade-offs at the failure boundary.

Background

Resistance-training-induced hypertrophy is driven, at the proximate physiological level, by mechanical tension imposed on motor units recruited during contraction. Higher-threshold motor units controlling type II muscle fibers are recruited progressively as a set continues and lower-threshold units fatigue, with full motor-unit recruitment achieved at or near voluntary failure for most exercise modalities. The implication is that working sets terminated well short of failure may not recruit the higher-threshold motor units that contribute disproportionately to hypertrophy, and conversely that the marginal hypertrophic value of the last few reps before failure may be substantial.

This article summarizes what the trial-grade and meta-analytic literature actually supports about how close working sets need to be taken to failure for hypertrophy, the role of RIR as a prescription tool, and the trade-offs at the failure boundary.

Defining the terms

Muscular failure in the resistance-training literature refers to the point during a set at which the trainee can no longer complete an additional concentric repetition with proper form. Reps in reserve (RIR) is the number of additional repetitions that could have been completed before reaching that point. A set taken to 2 RIR is one terminated when the trainee judges that two more reps could have been performed; a set taken to 0 RIR is taken to muscular failure.

Rate of perceived exertion (RPE) in the strength-training context, as operationalized by Helms and colleagues (2016), maps to RIR on a 1-to-10 scale: an RPE of 10 is muscular failure (0 RIR), an RPE of 9 is 1 RIR, an RPE of 8 is 2 RIR, and so on. The Helms RPE-RIR scale is now the most common load-prescription tool in evidence-based strength-training practice.

What the meta-analytic data show

The most comprehensive synthesis of this question is the Grgic et al. (2018) meta-analysis, which aggregated 15 controlled trials comparing failure and non-failure training protocols matched for volume and load. The pooled effect on muscle hypertrophy was small and not statistically significant, supporting the position that training to failure on every set is not necessary for hypertrophy when set count and load are equated. The same meta-analysis found a small advantage for failure training on muscular strength outcomes, although this finding was less robust than the hypertrophy null.

The Refalo et al. (2023) meta-analysis added trial data accumulated since 2018 and approached the question from a continuous perspective, modeling hypertrophy as a function of mean RIR rather than a binary failure/non-failure comparison. Their analysis supported a small advantage for training closer to failure (lower mean RIR) within the range of approximately 0 to 5 RIR, with hypertrophy declining more steeply as mean RIR rose above approximately 5. The shape of the curve is consistent with the motor-unit-recruitment account: sets terminated very far from failure may fail to recruit the higher-threshold units that contribute most to hypertrophy.

The Lacerda et al. (2020) controlled trial directly compared a failure protocol to a non-failure protocol matched for total volume in trained men over 14 weeks. Both groups showed similar hypertrophy, supporting the meta-analytic conclusion. The non-failure group required higher per-session set counts to match volume, illustrating a practical trade-off: training closer to failure produces more stimulus per set but fewer recoverable sets per session and per week.

RPE and RIR as load-prescription tools

The case for RIR-based prescription rests on a simple argument: a fixed percentage of one-repetition maximum (e.g., 75 percent of 1RM for 10 reps) does not produce the same proximity to failure across days, across trainees, or across exercises. Daily readiness varies; the load that produces a 2 RIR set on Monday may produce a 0 RIR set on Friday. Prescribing by RIR rather than by percentage allows the trainee to operate at a defined proximity to failure regardless of daily readiness.

The Helms et al. (2023) work on RPE versus percentage-based loading tested whether RPE-prescribed loads produce different hypertrophy and strength outcomes than percentage-prescribed loads in periodized programs matched for sets and repetitions. The trial provides evidence that RPE-prescribed loading is at least non-inferior to percentage-prescribed loading on the measured outcomes, supporting RPE-RIR as a defensible prescription tool. The accuracy of trainee RIR estimates improves with training experience; novices systematically underestimate RIR (perceiving themselves closer to failure than they are), and intermediate-to-advanced trainees converge on accurate estimates within the 0 to 4 RIR range.

The practical trade-offs at the failure boundary

Training closer to failure increases per-set stimulus but raises three costs:

1. Fatigue accumulation per session: sets taken to 0 RIR fatigue the trained muscle and the central nervous system more than sets taken to 2 to 4 RIR, plausibly limiting subsequent set quality.
2. Recovery cost between sessions: very high effort sessions extend the recovery interval required before the same muscle can be productively trained again, which interacts with the training-volume and training-frequency prescription.
3. Form degradation: the last few reps before failure are the most likely to involve technique compromise, particularly on compound free-weight movements, which raises injury risk and may shift load away from the target musculature.

For these reasons, evidence-based programming generally prescribes the majority of working sets at 1 to 4 RIR rather than uniformly at 0 RIR, with sets taken to outright failure reserved for selected exercises (typically isolation movements late in the session) and selected mesocycle phases (typically the final week of an accumulation block).

What the evidence does not support

The literature does not support training every set to failure as a hypertrophy maximizer; the meta-analytic evidence is consistent with comparable outcomes in the 0 to 4 RIR range. It does not support sets terminated more than approximately 5 reps short of failure as adequate for hypertrophy at typical volumes. It does not yet provide robust trial-grade evidence on the optimal mean RIR across long training timeframes (multi-year programs); the available data are from trials of 6 to 16 weeks, and the long-term cumulative trade-off between effort per set and weekly recoverable volume remains an open question.

The strongest defensible position is that working sets taken to within 0 to 4 reps in reserve produce comparable hypertrophy when volume is equated, that RPE-RIR-based load prescription is a more accurate load-targeting tool than fixed-percentage-of-1RM prescription, and that the marginal hypertrophic benefit of grinding to absolute failure on every set is small relative to the recovery cost.