REF-1609 · Wiedenmann et al., 2026 · BMC Sports Science, Medicine and Rehabilitation · Paper · PubMed · Free full text

Calling reps in reserve got better with practice: 26 adults with lifting experience but none with RIR (13 aged 33 ± 7 and 13 aged 67 ± 4) did three sets of bench press and leg press at 75-85% 1RM in six sessions over six weeks, calling 4 and 2 reps left and then going on to failure, and their mean absolute error fell by 0.4 to 2.3 reps from the first to the sixth session (most on leg press at 4 reps left), significantly below session 1 at sessions 4 and 6 in all four lift-and-target models and at session 5 in three of them, with no significant difference between the age groups; there was no group that did not practise. Read in full

Main caveat: No control group without practice; repeated 1RM testing before every session may itself teach where failure lies (inference).
limited evidenceMeasured on recreational peopleRead in full

Wiedenmann and colleagues asked whether estimating reps in reserve is a skill that improves with practice, and whether older adults judge it as well as younger ones. After a session that taught what concentric failure means and how to call it, 26 adults did six sessions over six weeks.

Each began with a 1RM test on leg press and bench press, followed by three sets of each lift at a load drawn at random between 75% and 85% of that day's 1RM. During each set they called out the rep at which they felt 4 reps and then 2 reps were left, and kept going to failure; no one told them how far off they were.

Mean absolute error fell across sessions for both lifts at both targets, with no significant age effect and no significant age-by-session interaction. Errors were smaller on bench press than leg press and at 2 than at 4 reps left, and grew with the number of reps a set allowed.

The authors suggest that taking every set to failure gave the lifters their own feedback, and that people who start out accurate, as in earlier studies of trained lifters, may have little room to improve.

Participants

30 recruited, 26 completed (two dropped out in each group: scheduling, prolonged illness, back pain unrelated to the study). Younger: n = 13 (9 women, 4 men), 33 ± 7 years (19-40).

Older: n = 13 (7 women, 6 men), 67 ± 4 years (60-74). 1RM bench press 37.5 ± 15.0 vs 29.5 ± 13.0 kg; 1RM leg press 132 ± 15 vs 96 ± 37 kg (only leg press 1RM differed significantly at baseline, p = 0.008). Recreationally active, with resistance-training experience, no experience with RIR or RPE regulation.

Protocol

Schedule: session 1 familiarisation (failure, verbal cues); sessions 2-7 identical, 5-10 days apart, under 60 min, one-to-one supervision.
Each session: warm-up; 1RM test for leg press, then bench press; three sets of each lift at a random 75% or 85% of that day's 1RM, at least 5 min rest between sets; real-time verbal call at 4 RIR and at 2 RIR, judged separately, then on to concentric failure.
Feedback: none given on the size of the error, within or after sessions.
Outcomes: directional error and mean absolute error (MAE) between true and estimated reps, averaged over the three sets per target and session.

Key Results

MAE, reps (session 1 → session 6)YoungerOlder
Bench press, 2 RIR1.6 → 0.81.2 → 0.8
Leg press, 2 RIR2.9 → 1.42.5 → 1.7
Bench press, 4 RIR2.4 → 1.52.1 → 1.5
Leg press, 4 RIR4.6 → 2.34.6 → 2.5
Mixed-model estimate, session 6 vs session 1Reps (95% CI)
Bench press, 2 RIR−0.79 (−1.24 to −0.35)
Bench press, 4 RIR−0.92 (−1.55 to −0.30)
Leg press, 2 RIR−1.56 (−2.37 to −0.76)
Leg press, 4 RIR−2.26 (−3.40 to −1.11)
Time course: error was significantly lower than in session 1 at sessions 4 and 6 in all four models and at session 5 in three (bench press at 4 RIR, session 5: p = 0.078, although the authors' text says from session 4 onward in all four), and also at session 2 for bench press at 2 RIR; the main effect of session was significant in each (p < 0.01).
Age: no significant group difference at any session and no significant session-by-group interaction; the authors say this is not evidence of equivalence.
Where calls were harder: MAE was lower on bench press than leg press (−1.42 reps, 95% CI −1.96 to −0.88) and at 2 than at 4 RIR (−1.04 reps, −1.30 to −0.78).
Set length: reps per set averaged about 11 on bench press and 13.3-14.6 on leg press; total reps to failure correlated with error, weakly on bench press at 2 RIR (r ≈ 0.19-0.26) and moderately elsewhere (r ≈ 0.42-0.58).
Direction: both groups underestimated the number of reps at both targets on both lifts (the authors' summary).

Practical Application

Calling reps left improved with practice in this study: in adults aged about 33 and 67 who were new to the method, six weekly sessions in which every set went on to failure, so each call was checked against what was really left, cut the error by 0.4 to 2.3 reps, with no significant age difference. That checked practice is the only approach in the base shown to make effort calls more accurate; with no group that did not practise, part of the gain may be growing comfort with hard sets rather than better judgement.
The calls were least accurate where the set allowed many reps and the call came far from failure (leg press, 4 left, error 4.6 reps at the start); a call of "2 left" on a heavier set was the most trustworthy.
No coach corrected the calls; the information came from finishing each set (the authors' reading). Inference, not tested here: on lifts where taking a set to failure is unsafe, the same check would need safeties or a spotter.
Experienced, already accurate lifters may gain little: trained men whose starting errors were, by these authors' account, about one rep or less did not improve over six weeks (REF-1610), which the authors read as a ceiling.
The authors suggest four to six sessions over several weeks of familiarisation before relying on RIR targets, and targets nearer failure (such as 2 RIR) for tighter control.
Inference, not tested here: whether rating a ride and then seeing the power or heart rate it produced trains endurance ratings the same way is unknown.

Limitations

No control group without practice; repeated 1RM testing before every session may itself teach where failure lies (inference).
13 per age group; power was calculated for the session effect only.
Lift order was fixed (leg press before bench press) in every session.
Error is counted in whole reps; a group change under one rep may mean nothing for an individual (the authors' point).
Every set went to failure, which is not how most people train; one session a week only.
Two lifts, both supported; no free squats, hinges or endurance work.
The authors report using ChatGPT to help write the analysis code and review language.
How this entry was checked and changed
Why this level

the rubric's uncontrolled pre–post line: the fall in error has no comparison group that did not practise, so learning to call reps cannot be separated from growing comfort with going to failure or from the 1RM test repeated before every session; 13 per age group; the sample size was set for the session effect only, so the age comparison is exploratory (the authors' statement)