Research · Training Load Monitoring

How Hard Was That, Really? What the Research Says About RPE as a Training Load Tool

By Muhammad Dzulhisham · MSc Sports Coaching, NSCA-CSCS

A national sports institute can tell you exactly how hard an athlete's session was before the athlete has finished stretching — heart-rate zone distribution, GPS-derived distance and high-speed running metres, an accelerometer-based player-load score, all logged automatically. A coach running a squad out of a converted warehouse, a school hall, or three different parks across Singapore in the same week has none of that. The honest question is whether the gap between those two setups is a gap in data quality, or just a gap in equipment budget. The research on rating of perceived exertion says it's closer to the second.

What session-RPE actually is

Rating of Perceived Exertion isn't a new idea. Swedish physiologist Gunnar Borg formalised it as a measurable psychophysical construct in 1982, building the category-ratio scales still used across exercise science today.¹ What turned a single intensity rating into an actual training-load monitoring tool was a modification introduced two decades later by American exercise physiologist Carl Foster and colleagues at the University of Wisconsin-La Crosse. Roughly 30 minutes after a session ends — long enough that a brutal last rep or an easy cool-down doesn't distort the athlete's read of the whole session — the athlete rates the entire session on a modified 0–10 scale running from 0 ("Rest") through 3 ("Moderate") and 5 ("Hard") to 10 ("Maximal"). That single number is multiplied by the session's duration in minutes.²

The result, session-RPE, is one arbitrary-unit figure per session. A 45-minute session rated 6/10 produces a load of 270 AU. A 90-minute session rated 3/10 also produces 270 AU — two genuinely different training days, quantified onto the same scale, from a question that costs nothing to ask and nothing to answer.

The validation research

Foster's own 2001 paper in the Journal of Strength and Conditioning Research was a direct test, not just a proposal.² Athletes performed steady-state and interval cycling, plus basketball practice, with each session quantified two ways at once: session-RPE, and an objective heart-rate-based method (Banister's TRIMP, built from time spent in individualised heart-rate zones). The paper's conclusion was that session-RPE is "a valid method of quantitating exercise training during a wide variety of types of exercise" — including the non-steady-state, intermittent work that heart-rate-only methods handle poorly.

That finding held up once other researchers went looking for it in real training environments rather than a lab. Impellizzeri and colleagues tracked 19 youth soccer players (mean age 17.6) across a full training cycle, comparing session-RPE against three separate heart-rate-derived load methods — Banister's, Edwards', and Lucia's TRIMP — and found strong, consistent correlations across training types, concluding the method "does not require particular expensive equipment and can be very useful and practical for coaches."³ A 2017 review in Frontiers in Neuroscience compiled the wider body of evidence that had accumulated by then across soccer, basketball, rugby, swimming, rowing, karate, taekwondo, tennis, and more: session-RPE's correlation with Banister's TRIMP ranged from about 0.50 up to 0.99 depending on sport and session type, with most individual and team sports comfortably above 0.70; correlations with Edwards' and Lucia's TRIMP followed a similar pattern.⁴ The same review found session-RPE tracking meaningfully with GPS-derived external load in team sports too — correlating at 0.83 with total distance and 0.60 with high-speed running in rugby, and at 0.76 with accelerometer-based player load in soccer.⁴

Resistance training is the case that matters most for a strength coach, since it's exactly the modality where a heart-rate strap tells you the least — HR responds slowly and inconsistently to short, high-force, intermittent effort. Sweet, Foster, and colleagues tested session-RPE directly against resistance exercise bouts at 50%, 70%, and 90% of 1RM, alongside aerobic bouts at matched intensities, in twenty adults, and found the method extended to lifting sessions as well as it did to cardio.⁵

The tool with essentially no cost and no equipment turns out to correlate with the expensive ones about as well as they correlate with each other.

What a subjective number catches that a strap doesn't

This isn't really a surprise once you consider what RPE is actually integrating. A heart-rate monitor measures cardiovascular strain. A GPS unit measures distance and speed. Neither one directly measures mechanical loading, eccentric stress, heat, sleep debt, or the cognitive and emotional cost of a tactically demanding session — all of which a trained athlete's brain is already weighing, consciously or not, when asked "how hard was that." The 2017 review's own summary of the field is that session-RPE functions as an integrated marker of global internal training load precisely because it isn't limited to one physiological channel the way a heart-rate strap or accelerometer is.⁴

None of this makes RPE flawless, and the same body of research is specific about where it gets noisier. Well-trained athletes tend to rate prescribed high-intensity work as harder, and prescribed low-intensity work as easier, than the coach intended — a real, replicated mismatch between planned and perceived training dose that is generally more pronounced in individual sports than team sports.⁴ Resistance-training sessions also show weaker correlations with heart-rate-based methods than steady aerobic or technical-skill sessions do, which is less a flaw in RPE than a reminder that heart rate was never a great reference standard for lifting in the first place.⁴ And the 30-minute collection window matters: rate it immediately and a hard finishing set skews the number; rate it the next day and the memory has already faded or been reshaped by how the athlete slept.

Turning single sessions into a monitoring system

A single session-RPE number is useful on its own. Summed across a week, it becomes something more — a monitoring system that predates GPS vests by decades. In 1998, Foster introduced two derived metrics still in routine use: monotony, the ratio of an athlete's mean daily training load to the day-to-day standard deviation of that load (a low number means genuinely varied hard and easy days; a high number means every day looks the same), and strain, the week's total load multiplied by its monotony score.⁶ Foster's original work tied elevated training monotony and strain to a higher incidence of illness in endurance athletes, and the field's working guidance since has been to treat sustained monotony above roughly 2.0 as a warning sign, with values below 1.5 preferred.⁶

The more recent version of the same idea is the acute:chronic workload ratio (ACWR) — comparing a short rolling window of session-RPE load (typically one week) against a longer rolling average (typically four weeks) to catch an athlete whose training has spiked well above what their recent training has prepared them for. A 2020 systematic review of 27 studies found a consistent trend toward the lowest injury risk sitting in an ACWR band of roughly 0.80 to 1.30, while also being direct that methodological weaknesses in how the ratio is calculated mean it should inform judgment rather than replace it.⁷ Both of these tools — monotony/strain and ACWR — run on nothing but the same duration-times-RPE number logged in a spreadsheet, week after week.

Setting it up week to week

None of the research above requires anything more sophisticated than a consistent routine, applied the same way every session:

• Ask at the same point every time. Roughly 30 minutes post-session — after the shower, before the memory fades or gets rewritten by how the evening goes — using Foster's original prompt: "How was your workout?"
• Use the same numeric scale, every athlete, every time. The modified 0–10 scale, with the word anchors ("Easy," "Moderate," "Hard," "Very Hard," "Maximal") visible so the number means the same thing to a 14-year-old and a 40-year-old on the same page.
• Multiply by session duration in minutes to get that day's load in arbitrary units, and log it — a shared spreadsheet is genuinely sufficient; the method was validated long before load-tracking apps existed.
• Sum the week to get weekly load, then track the trend rather than any single number in isolation — a spike relative to an athlete's own recent weeks matters more than the absolute figure.
• Build familiarity before trusting the number. Reliability improves with repeated exposure to the scale, which matters more for a new athlete — especially a younger one — than for someone who has rated a hundred sessions already.

Does it actually work with kids

This is where the evidence gets genuinely useful for a youth-focused programme, because it isn't uniform across ages and it comes with a real caveat worth taking seriously. A 2023 meta-analysis pooling 16 studies and 278 adolescent athletes found an overall correlation of r = 0.74 between session-RPE and heart-rate-based training load — "strong to very strong" in the authors' own classification — and concluded that session-RPE, on either the standard 0–10 (CR-10) or an expanded 0–100 (CR-100) scale, "can be used 'stand-alone' for monitoring internal training load for children and adolescent athletes."⁸ The CR-100 version performed somewhat better in that pooled data (r = 0.80 versus 0.69 for CR-10), though the authors were careful to note only three studies had tested it, too few to call it settled.⁸ For resistance training specifically in even younger children, the Children's OMNI-Resistance Exercise Scale — validated in 10-to-14-year-olds against actual weight lifted — showed concurrent validity coefficients of 0.72 to 0.88 for boys and girls alike.⁹

The caveat matters as much as the validation. A 2023 study of 11-to-13-year-olds during soccer training found children could rate perceived effort meaningfully — heart rate and cumulative training load together explained a real share of their RPE variance, and 13-year-olds rated noticeably more accurately than 11-year-olds — but when the same children were explicitly instructed to raise or lower their training intensity in a later session, their heart rate barely moved even though their reported RPE did.¹⁰ In the researchers' own words, the finding suggests children "may not be able to alter their intensities in response to instructions" the way an adult athlete can. Practically, that cuts a specific way: a child's post-session RPE is a genuinely useful, honest read of what actually happened to them physiologically. It is not yet a reliable dial a coach can ask them to turn up or down in real time. The coach still has to design the intensity; the RPE number tells you afterward whether the design landed the way you thought it did.

The practical case, without any lab equipment

None of the equipment-based alternatives to session-RPE are bad tools — heart-rate monitoring and GPS tracking are genuinely useful where they're available and well-fitted. But they were built for continuous, cardiovascular, outdoor, adult work, and every one of those qualifiers is a place they start to fail: a GPS unit has nothing meaningful to say about a resistance-training session, a heart-rate strap struggles to fit a ten-year-old properly and drifts with every seasonal illness, and neither travels well across a coach's actual week of a school hall, a five-a-side pitch, a home gym, and a Muay Thai pad session, all with different athletes on different budgets.

Session-RPE asks one question that works identically across every one of those settings, requires no charging, no calibration, no strap that has to fit correctly to mean anything, and produces a number a coach can compare across an entire squad using nothing more than a shared spreadsheet. That combination — validated against the equipment-heavy alternatives across dozens of sports and, per the more recent evidence, across age groups down into early adolescence — is why it remains the default monitoring layer for a programme that doesn't have force plates or a rack of GPS vests sitting in a cupboard. The absence of expensive hardware was never actually the barrier to real training-load monitoring. The barrier was assuming you needed it.