REF-1485 · Refalo et al., 2024 · Journal of Strength and Conditioning Research · Paper · PubMed

Can lifters call 1 or 3 reps left mid-set?

On the barbell bench at 75% 1RM, 24 trained lifters were off by 0.65 reps on average (SD 0.78), slightly underpredicting. Bench only: do not assume it for a squat or a station.

Main caveat: Abstract only. No individual error distribution, exact training years, or attendance details.
limited evidenceMeasured on recreational to trained athletes

Refalo and colleagues tested whether trained lifters can predict reps in reserve during a set, as evidence for using RIR to end sets. Twenty-four resistance-trained men (n = 12) and women (n = 12) did 2 experimental sessions, each with 2 sets of barbell bench press to momentary muscular failure at 75% 1RM, and said aloud when they believed they had reached 1 or 3 RIR.

Accuracy was the difference between predicted and actual RIR. Across all sets and sessions raw accuracy was −0.17 ± 1.00 and absolute accuracy 0.65 ± 0.78 reps. Accuracy was statistically equivalent, within ±1 rep, between 1-RIR and 3-RIR calls, between set 1 and 2, and between session 1 and 2, and showed no relation with sex, years of experience or relative bench strength.

The authors conclude RIR prescriptions may be used to inform proximity to failure on set termination. Our reading: good evidence for this one lift in this population. A mean of 0.65 reps with an SD of 0.78 means some calls miss by more than a rep. The study does not show that the same holds for a squat.

Protocol

Design: repeated-measures accuracy study; 2 sessions, 2 sets each, all sets to momentary muscular failure.
Who: n = 24 resistance-trained: 12 male, 12 female. Age and training years are not in the abstract.
Task: barbell bench press at 75% 1RM; verbal call at perceived 1 RIR or 3 RIR during the set, then continued to failure.
Measure: RIR accuracy = predicted minus actual, as raw (direction) and absolute (size); equivalence range ±1 repetition; no significance level set.
Funding and interests: not stated in the abstract record.

Key Results

MeasureValue
Raw RIR accuracy, all sets and sessions−0.17 ± 1.00 reps (minor tendency to underpredict)
Absolute RIR accuracy, all sets and sessions0.65 ± 0.78 reps
1-RIR vs 3-RIR callsstatistically equivalent within ±1 rep
Set 1 vs set 2statistically equivalent within ±1 rep
Session 1 vs session 2statistically equivalent within ±1 rep
Sex, years of experience, relative bench strengthno evidence of a relationship with accuracy

Mechanisms

Measured: the call against reps actually done.
Authors' interpretation: trained people can predict 1 and 3 RIR with high accuracy on the bench press, so RIR can inform set termination in research and practice.
Inference: every set went on to failure, so the feedback that real training lacks was on offer at each try; normal sessions that stop at the call never test it. Not tested in the abstract.

Practical Application

Can lifters call 1 or 3 reps left mid-set? On the barbell bench at 75% 1RM, 24 trained lifters were off by 0.65 reps on average (SD 0.78), slightly underpredicting. Bench only: do not assume it for a squat or a station.
Where it helps the guide: a "2+ reps to spare" check (SYNTH-44, practice) is plausible for trained lifters on the barbell bench press at a similar load (inference; only 1 and 3 reps left were called).
Close calls are ok, here: 1-RIR and 3-RIR calls were statistically equivalent within the authors' ±1-rep range, which is wide against a mean error of 0.65 reps. REF-1484, on other lifts, reports error of ≤1 rep up to 5 left on chest press and up to 3 on leg press, and over 2 at 7-10 left.
Not from this paper, and a suggestion for the new guide rather than a finding: for a lift or station with no data on accuracy, read reps, RPE and the trend in the log rather than trusting one call.

Limitations

Abstract only. No individual error distribution, exact training years, or attendance details.
One lift. Barbell bench at one load. The title and abstract limit it to this exercise; REF-1484 shows accuracy differs between exercises.
Sets to failure. Accuracy was obtained in sets continued to failure, with feedback each time (inference), in two sessions.
The equivalence margin is the authors'. ±1 rep is half of a 2-rep rule's margin, and an SD of 0.78-1.00 shows individual calls can fall outside it.
Small. 24 people, so "no evidence of a relationship" with sex or experience is weak evidence of none.
Not endurance athletes.
How this entry was checked and changed
Why this level

rubric's cross-sectional or validation study; n = 24, one exercise; a true reference (reps actually completed) and a sensible equivalence analysis, with the ±1-rep range chosen by the authors