For strength, stopping short costs you nothing. For muscle, closer to failure helps, and the last rep is the one with the least evidence behind it.
Training to failure means continuing a set until you cannot complete another repetition with the form you started with. Not the rep that felt hard. The rep that does not come.
The argument sits underneath almost every set you will ever write. If a set that stops two reps short builds roughly what a set to failure builds, then failure is a cost you are choosing to pay. If it does not, then stopping short is a discount you are choosing to take. Both are decisions, and most lifters make them by habit rather than on purpose.
Failure matters because it is the one point on the effort scale you can identify without a guess. Everything short of it is an estimate: you stopped, and you believe you had two reps left. That belief is what reps in reserve measures, and RPE, in its resistance-training form, is the same scale read from the other end. An RPE of 9 is one rep left. An RPE of 10 is none.
The scale is a skill, and it is trainable. Zourdos and colleagues built the RIR-based RPE scale specifically for resistance training, and their primary aim was to compare ratings between experienced and novice squatters. A later study by Ormsbee and colleagues took the same question to the bench press. Both treated training experience as a variable in how well people judge their own proximity to failure, which is a polite way of saying that a beginner's 2 RIR and a veteran's 2 RIR are not guaranteed to be the same number.
On strength, the research agrees. Vieira and colleagues in 2021 and Grgic, Schoenfeld and colleagues in 2022 found no significant difference in strength gains between training to failure and stopping short. When the two groups did not do the same volume, stopping short came out ahead, and in Vieira's analysis it did for power output as well. Robinson and colleagues, in a 2024 series of meta-regressions, found strength gains similar across a wide range of reps in reserve.
On muscle, the answer is a gradient, not a switch. Robinson's models found that muscle size increased as sets ended closer to failure. Refalo and colleagues, pooling fifteen studies in 2023, found only a trivial advantage for failure training, and no clear advantage at all when failure meant true momentary muscular failure. They suggest the relationship may not be a straight line. Read together: for muscle, closer to failure is better, and the final rep is the part with the least evidence behind it.
Two limits travel with those findings. First, several results change with volume. Vieira found failure ahead for growth overall, but not once both groups did the same volume. Second, highly trained lifters are under-studied. Grgic found a small advantage for failure in resistance-trained participants, but it is a subgroup result, and the authors call for more studies in highly trained people.
That makes proximity to failure a dose, like sets, and it travels with them. Ten sets left two reps short and ten sets taken to failure are not the same week of training.
So plan the ramp. Do not let it drift. If you are counting volume, a block that moves from RIR 3 to RIR 1 on purpose can be lined up against the next block that does the same, week for week. A block where every set quietly creeps toward failure cannot be compared with anything, and it tends to feel fine for three weeks and then does not.
The rest is practice rather than evidence, and it is labelled that way. Taking a heavy barbell compound to a true failure carries a cost the research on muscle growth does not price in: a missed squat is a safety event, and the fatigue lingers into the next lower-body session. Isolation work on a machine or a cable fails safely, recovers quickly, and is the usual place lifters choose to go all the way. Plenty of experienced lifters keep one or two reps in the tank on the lifts that matter and spend failure where it is cheap.
You are on seated cable rows, 3 x 10. Written as most lifters write it: three sets of ten, as hard as you can, which tells you nothing about how hard that was.
Written as a prescription: sets one and two at RIR 2, the last set to failure. Sets one and two end at ten reps with two honest ones in reserve. The last set is allowed to end wherever it ends: maybe eleven, maybe nine.
The log line now carries information. Next week's set one is compared to this week's set one at the same effort, and the failure set is a single deliberate data point instead of the whole week.
Prescribed as: 3 x 10 @ RIR 2, final set to failure.
Effort is stored with the prescription as a number, not as the word hard. Each week of a block carries its own RIR target, and that target is written into every set it prescribes. A set at RIR 2 is a different prescription from a set at RIR 0, and the number travels with the set rather than living in a plan document you read three weeks ago.
Plans built in Fire Your Coach ramp that target on purpose, from more reps in reserve early in a block to fewer late, and reset it at the deload. Week three of this block lines up with week three of the next.
When you log a set, you can record the RPE it actually took. The target sits on the prescription and the actual sits on the set, so the gap between them is visible. A volume jump that came from creeping toward failure looks different from one that came from getting stronger.
The papers behind the empirical claims on this page, with the population each one studied. Where we could confirm only what a study set out to measure and not the size of what it found, the note says so rather than borrowing the paper's authority for a number nobody read.
Alexandra F Vieira et al. (2021). Effects of Resistance Training Performed to Failure or Not to Failure on Muscle Strength, Hypertrophy, and Power Output: A Systematic Review With Meta-Analysis. Journal of Strength and Conditioning Research.
No difference in maximal strength overall (SMD -0.08, p = 0.642); non-failure produced greater strength gains when volumes were not equalised (SMD -0.34, p = 0.048). Failure produced greater hypertrophy overall (SMD 0.75, p = 0.005), with no difference when volumes were equalised. No difference in power overall (SMD -0.20, p = 0.239); non-failure better for power when volumes were not equalised (SMD -0.61, p = 0.025).
Systematic review with meta-analysis of training to failure versus not to failure; results reported separately for studies with equalised and non-equalised training volume.Jozo Grgic 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.
No significant difference between failure and non-failure for strength (ES -0.09, 95% CI -0.22 to 0.05) or hypertrophy (ES 0.22, 95% CI -0.11 to 0.55). Non-failure favoured for strength when volume was not equated (ES -0.32, 95% CI -0.57 to -0.07). In the resistance-trained subgroup, failure favoured for hypertrophy (ES 0.15, 95% CI 0.03 to 0.26).
Systematic review and meta-analysis; subgroups by training status and by whether volume was equated. The authors call for more studies in older adults and highly trained individuals.Martin C Refalo et al. (2023). Influence of Resistance Training Proximity-to-Failure on Skeletal Muscle Hypertrophy: A Systematic Review with Meta-analysis. Sports Medicine.
Trivial advantage for training to set failure versus non-failure for hypertrophy across any failure definition (ES 0.19, 95% CI 0.00 to 0.37, p = 0.045), with no moderating effect of volume load or relative load. No advantage for momentary muscular failure specifically (ES 0.12, 95% CI -0.13 to 0.37, p = 0.343). The authors conclude there is no evidence that training to momentary muscular failure is superior for hypertrophy, and suggest a potentially non-linear relationship.
Systematic review; searches of PubMed, SCOPUS and SPORTDiscus identified 15 studies.Zac P Robinson 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.
Estimated RIR had a negligible relationship with strength gains in all best-fit models. For hypertrophy, the slopes were negative with confidence intervals excluding null: muscle size increased as sets ended closer to failure. Conclusion: strength gains were similar across a wide range of RIR, while hypertrophy improves as sets are terminated closer to failure.
Series of meta-regressions on estimated proximity to failure (RIR) across the existing literature.Michael C Zourdos et al. (2016). Novel Resistance Training-Specific Rating of Perceived Exertion Scale Measuring Repetitions in Reserve. Journal of Strength and Conditioning Research.
The stated primary aim was to compare RPE values measuring repetitions in reserve at particular percentages of 1RM in experienced versus novice squatters.
29 individuals, mean age 24.0 +/- 3.4 years, experienced and novice squatters.Michael J Ormsbee et al. (2019). Efficacy of the Repetitions in Reserve-Based Rating of Perceived Exertion for the Bench Press in Experienced and Novice Benchers. Journal of Strength and Conditioning Research 33(2): 337-345.
Extends the RIR-based RPE question from the squat to the bench press, again splitting experienced from novice lifters. Journal, volume and page range are self-cited in the abstract.
Experienced and novice male bench pressers, university setting.No. Sets that end closer to failure do build more muscle, but a 2023 meta-analysis of fifteen studies by Refalo and colleagues found only a trivial advantage for failure training, and no clear advantage for true momentary failure. Train close to failure on the sets meant for growth. Going all the way is optional.
No. Meta-analyses by Vieira and colleagues in 2021 and Grgic and colleagues in 2022 found no significant strength difference between failure and non-failure training, and stopping short did better when volume was not matched. Robinson and colleagues in 2024 found strength gains similar across a wide range of reps in reserve.
Failure is the point where the next rep does not complete. Reps in reserve counts how many reps were left when you stopped. RPE in its resistance-training form is the same scale expressed the other way: RPE 10 is no reps left, RPE 9 is one, RPE 8 is two. Zourdos and colleagues built that scale for lifting specifically.
Less accurate than it feels, especially early on. Zourdos and colleagues compared RIR-based ratings in experienced and novice squatters, and Ormsbee and colleagues did the same on the bench press, so experience is treated as a variable. We did not read the size of the difference. The usable habit is to take an occasional set to true failure on a safe exercise and check what your 2 RIR actually meant.
As a deliberate choice, on exercises where it is safe and cheap to recover from: final sets of machine and cable work, mostly, and usually late in a block. That is practice rather than evidence. Keep a rep or two in reserve on heavy barbell compounds, and plan how your RIR targets move across a block so your set counts stay comparable.
Describe what you are training for and Fire Your Coach builds the block, carries the prescription onto the set, and remembers what the method actually produced.