Proximity-to-Failure Training: What the Meta-Analysis Actually Found

For about forty years the dominant message in strength training was train harder. Train to failure, past failure, through drop sets, forced reps, rest-pause — the harder the set, the better the result. In 2023 that message ran into a meta-analysis it couldn't talk around. Refalo and colleagues pooled the trials that had directly tested training to failure against training stopped 1–3 reps short, and the answer was uncomfortable for anyone still selling intensity theatrics: the two approaches produce roughly equivalent hypertrophy, and failure costs more.

What Refalo 2023 tested

The paper — Influence of Resistance Training Proximity-to-Failure on Skeletal Muscle Hypertrophy: A Systematic Review with Meta-analysis, published in Sports Medicine — synthesised 15 studies (roughly 350 participants) comparing training conditions that differed only in how close each set was taken to muscular failure. The conditions clustered into three buckets:

• Failure (RIR 0) — sets terminated when another rep was impossible
• Close to failure (RIR 1–3) — sets terminated deliberately 1–3 reps shy
• Further from failure (RIR 4+) — sets terminated well short

The primary outcome was site-specific muscle hypertrophy (measured by ultrasound, MRI, or DXA depending on study). The result: no statistically meaningful difference in hypertrophy between failure training and close-to-failure training when volume was equated. Training further from failure (RIR 4+) did produce less hypertrophy — stopping that early genuinely costs you gains.

Why the fatigue cost matters

If failure and RIR 1–3 produce the same growth, failure looks equivalent. It isn't, because the secondary outcomes tell a different story:

• Recovery time — failure sets require longer inter-session recovery. Participants in failure conditions often trained fewer total sessions over the study period at matched intended volumes because they couldn't recover on schedule.
• Perceived exertion and session RPE — consistently higher for failure conditions. Translation: training felt worse without producing better results.
• Set-to-set performance degradation — later sets in a failure-to-failure session showed steeper rep drop-off, meaning the "total reps performed" advantage of going hard on every set evaporates across a multi-set workout.
• Joint and form issues — qualitative reports, not meta-analysable, but consistent.

Put those together and the practical implication flips: for the same hypertrophy outcome, RIR 1–3 lets you train more frequently, with better form, at lower subjective cost. That's a clear win.

What "equivalent" doesn't mean

Worth stating clearly, because the result gets misread both ways:

• It does NOT mean effort doesn't matter. Training at RIR 4+ produced less hypertrophy than training at RIR 0–3. You still need to be close to failure. You just don't need to cross into it.
• It does NOT mean failure is "bad" or you'll lose muscle by pushing. The effect is small and the confidence intervals on failure vs RIR 2 overlap. The cost-benefit analysis, not the growth signal, is what tips the call.
• It does NOT generalise to every context. The trials pooled were mostly compound lifts in a gym context with trained populations. Contexts where failure might remain defensible — final working set on an isolation, AMRAP test sets, final week of a mesocycle — weren't the focus.

The coaches' consensus

The major evidence-based programmers have converged on roughly the same position:

• Mike Israetel / RP — RIR progressions through a mesocycle, failure reserved for the final week and mostly on isolations
• Jeff Nippard — RIR 0–2 for most working sets on compounds, RIR 0 accepted on isolations, heavy use of RIR to regulate daily intensity
• Eric Helms — auto-regulation via RPE/RIR, choosing effort day by day rather than by block schedule

The disagreement between them is about how often and where to cross into RIR 0, not about whether RIR 1–3 is the sensible default. All three would nod at Refalo 2023 as formalising what their programming had been doing for years.

Programming it on a cut

The failure-vs-RIR calculation changes on a cut for the same reasons training volume on a cut does: recovery capacity drops, so the fatigue cost of each failure set is higher in relative terms. Roth 2022's narrative review on training during energy restriction reaches the same conclusion Refalo does — RIR 1–3 is the defensible default — and adds that even RIR 0 on isolations is worth reconsidering as calories drop.

Practical rule for a cut:

1. RIR 1–3 on all compound work, every session, every week
2. RIR 1–2 on primary isolation work in the first half of a block
3. RIR 0 only on the last set of an isolation, in the last week of a block, before a deload — if at all

If you're running a 12+ week cut (fat-loss phase), err toward the upper end of those RIR ranges. The marginal growth gain from pushing closer to failure does not survive the recovery tax.

When failure still earns its keep

The evidence doesn't say "never train to failure." It says "failure is not a prerequisite, and it costs more than it adds." Defensible use cases:

• The last set of an isolation exercise — low systemic cost, the growth stimulus/fatigue ratio is favourable
• Single-set "challenge" protocols for lifters short on time (the one-set-to-failure research supports hypertrophy from single max-effort sets)
• End-of-block AMRAP tests for measurable progression tracking
• Deload-adjacent "finishing" weeks before a recovery block

Outside those cases, the meta-analysis is clear: stopping 1–3 reps shy is not compromise, it's programming. The results are the same; the costs are lower. That's a rare clean win in strength science.

Key Takeaways

• Refalo 2023 meta-analysis pooled 15 trials; RIR 0 (failure) and RIR 1–3 produce equivalent hypertrophy at equated volume
• Failure costs more recovery time, higher RPE, worse form, and greater injury risk per unit of growth
• Training at RIR 4+ does produce less hypertrophy — you still need to be close to failure, just not crossing it
• Coach consensus (Israetel, Nippard, Helms) is RIR 1–3 as default, with failure reserved for selective contexts
• On a cut, default to RIR 1–3 on all work; reserve RIR 0 for the last set of a block on isolations only
• Compound lifts (squats, deadlifts, bench, rows) stay at RIR 1–2 minimum throughout a cut

Sources

1. Refalo MC, Helms ER, Trexler ET, Hamilton DL, Fyfe JJ (2023). Influence of Resistance Training Proximity-to-Failure on Skeletal Muscle Hypertrophy: A Systematic Review with Meta-analysis. Sports Medicine. PubMed
2. Grgic J, Schoenfeld BJ, Orazem J, Sabol F (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. PubMed
3. Schoenfeld BJ, Grgic J (2021). Effects of Range of Motion on Muscle Development During Resistance Training Interventions: A Systematic Review. SAGE Open Medicine. PubMed
4. Roth C et al. (2022). Training Volume and Intensity during Energy Restriction: A Narrative Review. Sports Medicine — Open. PMC