Afitpilot®
← Back to Glossary
Intensity & Effort

RPE (Rate of Perceived Exertion)

Also known as: Perceived Effort, Subjective Intensity

A 1-10 self-reported scale measuring how hard a set felt. In strength training, RPE is anchored to "reps in reserve" (RIR): RPE 10 means you couldn't do another rep, RPE 8 means you had roughly 2 reps left, RPE 6 means about 4 reps in reserve.

Formula

Session RPE = sum(exercise RPE x sets) / sum(sets) [weighted average across all exercises with RPE data, minimum 2 exercises required]

Example

Squat: RPE 8 x 4 sets = 32. Bench: RPE 7 x 3 sets = 21. Rows: RPE 7.5 x 3 sets = 22.5. Session RPE = 75.5 / 10 = 7.55, displayed as 7.6.

How Afitpilot Uses This

We compute a weighted session RPE from per-exercise RPE data, compare it to the target RPE from the prescription, and show the delta. An effort delta of +/-0.7 RPE is flagged amber, +/-1.2 is flagged red. RPE trends are tracked over your last 6 sessions using linear regression to detect "effort creep" (sessions getting harder than prescribed) or "under-reaching" (consistently easier than planned).

RPE accuracy by experience level

Who / ContextValueNote
Beginner (< 6 months)Off by 2-3 RPE pointsThink they're at RPE 7, actually RPE 4-5
Intermediate (1-3 years)Off by ~1 RPE pointReliable enough for programming
Elite powerlifterWithin 0.5 RPERPE is a trained skill, not just a number
Active aging (60+)Wider daily swingsSleep/stress affects RPE more — this is RPE working as designed
Runner's equivalentConversational = RPE 4-5, Tempo = 7-8, Race = 9-10Borg scale was invented for cardio first

Known Limitations

  • •RPE is subjective — it's influenced by sleep, stress, caffeine, mood, and individual calibration. Two athletes reporting RPE 8 may be at very different proximity to failure.
  • •We require a minimum of 2 exercises with RPE data to compute session RPE. With only 1 exercise reporting, the session RPE is null rather than potentially misleading.
  • •When an athlete marks a set as "completed as planned" without entering RPE, we substitute the target RPE as actual RPE. This is an optimistic assumption — the athlete may have found it harder or easier than prescribed. This proxy is noted in the UI but is a known source of inaccuracy.
  • •Mobility and warm-up exercises are excluded from the RPE calculation, which is correct for effort tracking but means those exercises contribute zero to the session effort score.

What We're Improving

  • Future

    Effective Sets (Hard Sets) Tracking

    Track sets taken within 0-3 RIR of failure separately from total sets. Research increasingly suggests that "hard sets" are a better predictor of hypertrophy stimulus than total volume. This would give a more nuanced view of training dose.

  • Future

    Prilepin's Table as LLM Strength-Prescription Guardrail

    Validate LLM-generated strength sessions against Prilepin's reps-per-set / total-reps ranges for the implied %1RM (derived from targetRpe → RIR → e1RM). Two surfaces: a plan-generation guardrail that flags or corrects out-of-range prescriptions, and a post-session signal that compares actual sets × reps × estimated %1RM against the optimal range. Deferred until we have 20+ real LLM-generated weeks audited; a ~8% high-intensity overshoot rate on a small fixture set was borderline-but-too-small to act on.

  • Shipped

    Universal Internal-Load Currency (sRPE → AU)

    Move from tonnage-as-primary to a universal session-load number using Foster's sRPE × duration method. AU is now the headline load metric on every session card and rolls up weekly with a 3-bucket modality split (Strength / Endurance / Mixed). Tonnage stays visible as a strength-specific secondary.

Science Context

The RPE scale used in resistance training is based on Borg's original CR-10 scale, adapted by Helms et al. (2016) and Zourdos et al. (2016) for strength training. Research shows RPE correlates well with actual reps-in-reserve for trained lifters, but novices tend to underestimate proximity to failure by 2-3 reps. RPE reliability improves with training experience.