How Close to Failure Should You Train? The RIR Evidence

The shape of the curve

Two claims, and people usually only hear the first.

Claim one: failure is not required. A set stopped at 2 RIR produces hypertrophy comparable to the same set taken to 0 RIR. This is the finding that gets quoted.

Claim two: there is a floor, and it is closer than people think. Above about 5 RIR the stimulus falls off steeply. A set at 6 reps in reserve is not a slightly less effective working set — it is functionally a warm-up.

The consequence is a usable band rather than a target:

The velocity-loss analogue

The same relationship shows up in the velocity-loss literature, which is useful because it is a measured quantity rather than a self-report.

Velocity loss above 25% within a set beats loss under 20%, and is no better than 20 to 25%. That maps closely onto the RIR picture: there is a threshold below which the stimulus is insufficient, and past it, going further buys nothing.

Two independent measurement approaches agreeing is more convincing than either alone.

The prescription

Every working set in 0 to 4 RIR, most in 1 to 3, with a split by movement type:

Compounds — squat, hinge, press, row: 2 to 3 RIR. Failure here costs disproportionate systemic fatigue for no extra hypertrophy. A missed rep on a heavy squat is a hole you spend a week climbing out of, and the stimulus you bought was the same as a set at 2 RIR.

Isolation — curls, lateral raises, extensions, leg curls: 0 to 1 RIR. Cheap to recover from, and the last reps are where the stimulus actually lives. There is no systemic cost to failing a lateral raise.

Why this matters more for hybrid athletes

The systemic fatigue from a set taken to failure has a second claimant.

For a pure lifter, the cost of training to failure on compounds is that the next lifting session is worse. For someone who also has a threshold run on Thursday, the cost is that the run is worse too — and the run being worse gets attributed to interference rather than to Tuesday's last set of squats.

Two to three RIR on compounds is not conservatism. It is what leaves a heavy lower-body session recoverable alongside a hard endurance session, and it costs nothing in hypertrophy terms.

RIR as a fatigue signal

The other reason to record it, and the one that gets missed: RIR at a fixed load is the clearest fatigue signal available.

If a weight you handled at 2 RIR three weeks ago now comes out at 0 RIR, your fitness has not fallen. Fatigue has accumulated to the point where it is masking fitness. That is the textbook deload signal, and it is completely invisible in a log that records only weight and reps.

It is one of four signals in a reactive deload trigger — alongside depressed recovery against your own baseline, three or more short nights in a week, and a flat strength trend. Two firing at once is a pattern. How the trigger works.

RIR and the progression rule

Without RIR, double progression cannot distinguish two opposite situations:

Both look like "hit the top of the range" to a rep-only rule, which will add five pounds to both. For the second case that is months of slow-walking through weights you could already handle.

The rule with the clause:

every set at top of rep range AND RIR <= 1  ->  add load, reset to bottom of range
inside the rep range                        ->  hold load, add one rep
below bottom of range, or RIR > 4           ->  hold; twice in a row, drop 10%

How accurate is self-reported RIR

Imperfect, and it improves substantially with practice.

The consistent finding is that lifters — trained ones included — underestimate how many reps they have left more often than they overestimate, particularly on lower-body compounds and particularly early in a training career. A reported 2 RIR is frequently a true 4.

Two things follow. First, this systematically works against you: you think you are training in the productive band and you are at the edge of it. Second, the fix is calibration — occasionally taking an isolation set genuinely to failure and comparing it to what you predicted.

Despite the error, it remains the most useful field measure available. No wearable measures proximity to failure, velocity devices are expensive and impractical for most people, and the alternative — not recording it — is strictly worse than an imperfect estimate.

What to do next

For four weeks, record RIR on every working set and hold your loads. Then look at whether the number drifts upward at a fixed weight. That drift, or its absence, tells you more about whether your block is working than any single session ever will.