FormaGet FormaRIR is useful, not perfectly accurate.
Repetitions in reserve is a prediction of remaining capacity. Research finds meaningful error, changing accuracy across loads and proximities to failure, and value in individualized calibration.
Published and last reviewed September 14, 2026 · 8-minute read
What the review found
| Factor | Observed tendency | Interpretation |
|---|---|---|
| Average error | Participants underpredicted task-failure repetitions by 0.95 on average. | A reported 2 RIR should not be read as an objectively proven two. |
| Proximity to failure | Predictions tended to improve closer to failure. | Estimating 2 RIR may be easier than estimating a large reserve. |
| Relative load | Heavier loads tended to improve prediction. | High-repetition sets can create more room for counting error. |
| Training status | Background did not clearly explain accuracy. | Experience alone does not guarantee precise estimates. |
What newer studies add
In 36 resistance-trained men and women performing bench press, individualized velocity measures were more accurate than subjective estimates in several initial-set and lower-load comparisons at 4 and 2 RIR. A separate squat study in 46 trained people found acceptable prediction from individualized, but not general, RIR-velocity relationships, with mean error below two repetitions in the subsequent session.
A 2025 study in 25 community-dwelling older adults found underprediction at intended 2 and 4 RIR in its chest-press protocol and limited separation between reported RPE values. The authors described RIR as lacking precision for exercise prescription in that sample while potentially retaining value for volume monitoring.
Technology does not remove context
Velocity relationships require equipment, exercise-specific testing, and individualized data. These studies do not make one velocity threshold or general model universally accurate.
A practical calibration process
Standardize
Keep exercise, range, technique, and equipment comparable.
Predict
Record the RIR estimate immediately when the planned set ends.
Check selectively
Occasionally continue only when the exercise and setting make approaching failure appropriate and safe.
Review error
Compare predicted and actual reserve across several sessions instead of trusting one test.
Calibration is a learning process, not proof that every future rating will be exact. Pain, unsafe setup, or technique breakdown should stop a set regardless of the planned check.
How to log RIR responsibly
Keep the raw performance
Exercise, load, repetitions, range, and working sets remain the observable record.
Add RIR as context
The rating helps explain perceived effort, but should not erase uncertainty or prescribe a load change by itself.
Frequently asked questions
How accurate is RIR?
RIR is useful but imperfect. A review of 12 studies found that participants underpredicted repetitions to task failure by 0.95 repetitions on average, with substantial variation between conditions and people.
When is RIR more accurate?
The review found estimates tended to improve closer to failure, with heavier relative loads, and in lower-repetition sets. These are group-level tendencies, not guarantees for one lifter or exercise.
Can velocity measure RIR objectively?
Individualized velocity profiles improved prediction in some bench-press and squat studies, but they still produced error and require exercise-specific testing and equipment. A general profile should not be treated as exact.
How can I calibrate my RIR estimate?
Keep the exercise and technique stable, record RIR immediately, compare repeated sessions, and occasionally test predicted reserve only in an exercise and setting where approaching failure is appropriate and safe.
Sources
Scoping review and exploratory meta-analysis, 2022. PMID: 34542869.
Bench-press study, 2024. PMID: 38595310.
Squat study, 2024. PMID: 38418370.
Chest-press validation study, 2025. PMID: 40902458.
Continue through the RIR library
Keep the estimate beside the evidence.
Forma records optional RIR with the observable set history. It does not present the rating as a direct measurement or publish its proprietary progression logic.
Use the calculatorDownload Forma