If you train by feel, you're leaving progress on the table. Loads that feel "heavy" today feel easy tomorrow, and vice versa. Without an internal yardstick, you push too hard on the wrong days and too little when you could be raising the bar.
RPE is the first serious tool to stop guessing: a scale validated in the literature that translates perceived exertion into a number, and that number becomes the key to picking the load, managing fatigue and building real progression. Let's see how to use it, not how to quote it.
1. What RPE Is — No-Nonsense Version
RPE stands for Rate of Perceived Exertion. In the gym, in its modern form, it's a 1-10 scale that measures how hard a set is in relation to muscular failure.
Operating rule: RPE 10 = technical failure, you can't do another clean rep. From there you scale down: RPE 9 = one rep in reserve, RPE 8 = two in reserve, and so on.
RPE doesn't measure how much your muscles burn, how "heavy" the load is in absolute terms, or how mentally tired you are. It measures one thing only: how many reps you could still have done with good technique before stopping. If you could have done three more at the end, that's RPE 7. Period. Operating numbers that serve programming, not feelings.
2. Where the Scale Comes From
RPE's history doesn't start in the gym, but in the lab. In 1962 the Swedish psychophysiologist Gunnar Borg proposed the Borg 6-20 scale, designed for perceived exertion in cardiovascular work. The "odd" numbers from 6 to 20 are aligned with expected heart rate multiplied by ten (RPE 13 = roughly 130 bpm in a healthy young adult). It's the scale you still find today in endurance studies.
For lifting, though, 6-20 is awkward. In the 2000s the powerlifter and coach Mike Tuchscherer adapted it for the weight room: the 1-10 RPE scale based on RIR (Reps In Reserve) was born. Tuchscherer spread it inside his Reactive Training Systems around 2008, and from there it entered the programs of anyone serious about strength.
The leap to scientific legitimacy came in 2016 with the work of Mike Zourdos and colleagues in the Journal of Strength and Conditioning Research: the RIR-based scale was validated in trained lifters, and it was shown that in experienced athletes RPE is a reliable tool for monitoring real set intensity. From there the autoregulation literature exploded, and figures like Eric Helms and the Stronger By Science team (with Greg Nuckols) built entire programming models on top of RPE.
Three dates to remember: 1962 Borg, 2008 Tuchscherer, 2016 Zourdos.
3. The Modern RPE Scale
Here's the version you use in the gym. Memorise it, because the rest of the article revolves around it.
Half Units Matter
Half units (8.5, 9.5) aren't fussiness: they recognise real uncertainty. If you finish with "one for sure, maybe two", that's RPE 8.5. Forcing yourself into whole categories makes you fake a precision you don't have.
Below RPE 6, the Scale Loses Meaning
RIR beyond four are unreliable even for experienced athletes. In practice you use it operationally from 6 to 10. Below that it's "light".
Technical Failure, Not Absolute Concentric Failure
RPE 10 = technical failure. If the rep gets ugly, slow, with cheating, you're already at RPE 10 even if in theory you could grind one more out with body english. The scale assumes technical quality.
4. The Truth About the Learning Curve
Here comes the part that 90% of articles on RPE skip entirely, and which is actually the most important.
RPE doesn't work right away. The first 4-6 weeks you'll misjudge it systematically. Research on less trained athletes confirms it: those who have never taken sets close to failure tend to drastically overestimate RIR. You think you're at RPE 8 and you're at 9.5, or you think you're at RPE 10 and you had three more in the tank.
The reason is simple: to estimate the reps you have left, you need to know what failure feels like. If you've never actually been there, you have no internal reference point. It's like asking someone who has never run a marathon at what kilometre they'd blow up.
How to Actually Calibrate Yourself
Three tools, in order of impact:
1. Film high-RPE sets. You thought RPE 9, you watch yourself slowing down dramatically on the last two reps with the bar zigzagging: it was RPE 10. The mirror lies, the camera doesn't. Bar speed is the objective proxy for effort.
2. Failure test every 4-8 weeks. On a safe exercise (leg press, lat pulldown, bench with safeties), take a set to true technical failure. The gap between actual reps and predicted reps at RPE 10 is your calibration error.
3. Benchmark weights. Periodically do 3 reps with a known load (e.g. your 5RM) and log the RPE. If over time the same weight always gives the same RPE, you're calibrated. If it bounces around randomly, you still have work to do.
Realistic Expectations
- 0-4 weeks: error of 1-2 full RPE points.
- 1-3 months: average error under 1 point on compounds.
- 6+ months of deliberate practice: error of 0.5 points on compounds (the level of athletes in the Zourdos study).
Expecting surgical precision from week one is the number one reason people drop RPE. It's a skill: it's trained, measured, improved.
5. RPE by Goal
There's no single "right RPE". The range changes based on what you're training.
Pure Strength (RPE 8-10)
Working sets at RPE 8-9, with occasional top sets at RPE 9.5-10 in intensification phases. Chronic RPE 10 doesn't work: fatigue accumulates faster than you recover. Evidence-based powerlifting programs (Stronger By Science, Helms templates) keep the bulk at RPE 7-9 and use RPE 9.5-10 sparingly, in the final weeks before a peak.
Hypertrophy (RPE 7-9)
The main stimulus is volume close to failure. Recent research confirms that sets at 0-4 RIR produce equivalent hypertrophy, provided volume is sufficient. Working in RPE 7-9 lets you accumulate more weekly volume than always being at RPE 10, and that volume is the real driver of growth. Failure is fine on isolation finishers; systematic failure on compounds burns the nervous system and cuts sustainable frequency.
Muscular Endurance (RPE 6-8)
High reps (15-30+) for metabolic work, conditioning, finishers. RPE 6-8 is the sweet spot: enough stimulus without neural fatigue. RPE 9-10 at high reps is no man's land: brutally taxing, slow recovery, marginal return.
Deload (RPE 5-7)
Volume and intensity dropped, loads at 50-65% of normal. A deload isn't "not training": it's keeping the motor pattern and letting fatigue dissipate.
RPE isn't a goal, it's an intensity constraint: you use it to stay in the right range for the target stimulus, depending on the phase.
6. RPE by Exercise
Here's another point most guides ignore: RPE reliability isn't equal across all exercises. Knowing when to trust the number and when to supplement it is what separates practical use from academic use.
Multi-Joint Compounds: High Reliability
Squat, bench, deadlift, military press, barbell row. RPE works well because failure is sharp (the bar slows down measurably) and the pattern is stable (movements repeated hundreds of times sharpen internal sensitivity). On compounds RPE is the primary load-selection constraint — "bench 4x5 at RPE 8" is a clean instruction.
Single-Joint Isolations: Low Reliability
Barbell curl, lateral raises, leg curl, leg extension, cable flyes. Here RPE loses precision because failure on isolations is often not muscular, it's leverage-based: on lateral raises you finish the set because you can't lift in line anymore, not because the deltoid is actually spent. The "one more for sure" sensation lies.
Practical prescription:
- Treat every isolation set as "around RPE 9-10 at the end"
- Prioritise the rep target (e.g. 12-15 reps) taken to technical failure, not a precise RPE
- Cluster, drop set, rest-pause as techniques, not micro RPE adjustments
On compounds you use RPE as a load-selection variable, on isolations you use reps to failure (or close to it).
Bodyweight: Systematic Overestimation
Push-ups, dips, pull-ups, ring work. Here RPE is treacherous for a simple reason: you can't change the load easily, and almost everyone overestimates RIR because the frustration of "it won't go up" arrives before real muscular failure.
Strategy: count AMRAPs and use the speed of the last rep as an indicator. If the last one came up slow and ugly, you were already at RPE 10. If you think you can do 3 more clean ones, you probably have 1-2.
Cardio: Use the Original Borg Scale, Not 1-10
Common mistake: applying the 1-10 RPE scale to cardio. It doesn't work, because 1-10 is anchored to muscular failure, while in cardio "max effort" is cardio-respiratory + lactate. For cardio use the Borg 6-20 scale:
- Borg 9-11: warm-up, recovery
- Borg 12-14: moderate aerobic zone (Z2)
- Borg 15-17: threshold, moderate intervals
- Borg 18-20: max effort, sprints
It's literally the tool Borg invented the scale for. Keep them separate.
7. A Full Session With RPE — Operational Example
Let's see how this translates into a real workout. Full body session, intermediate lifter.
A) Squat: 4x5 at RPE 8
Target: 5 reps with 2 RIR. Starting load 110 kg.
- Set 1: 110 kg x5. If you feel 3 in the tank -> RPE 7, go up to 115 kg.
- 2 in the tank -> RPE 8, stay at 110 kg.
- 0-1 in the tank -> RPE 9-10, drop to 105 kg.
B) Flat Bench: 4x6 at RPE 8
- Load: 80 kg x6.
- Same principle as the squat. Bar slowing noticeably on the last 2 reps? You were at RPE 8.5, drop 2.5 kg.
C) Barbell Row: 4x8 at RPE 7-8
Failure is often lower-back before lats.
- Load: 70 kg x8.
- If your back tilts vertical (loss of ROM) before 8 reps, you're already RPE 9-10. Drop the load.
D) Dumbbell Curl: 3x12 to Technical Failure
Isolation -> ignore precise RPE, form failure. Load: 14 kg/arm. Stop when you start swinging with your back.
E) Seated Leg Curl: 3x15 at RPE 9
Metabolic finisher, one rep in the tank is enough. Load: full stack -2.
Decision Flow After the First Working Set
- First set → estimate RPE
- RPE < target → go up (+2.5 kg compound, +1 kg dumbbells)
- RPE = target → hold
- RPE > target → go down (-2.5/-5 kg compound, -1/-2 kg dumbbells)
The first set is always the calibrator. The following ones are guided by that estimate.
8. From RPE to Programming: The Real Point of Autoregulation
Logging the RPE of a single set isn't worth much on its own. The value explodes when you connect today's numbers to tomorrow's decisions. This is autoregulation, the leap from "tracker" to "programmer" of your training.
The Basic Logic
You prescribed squat 4x5 at RPE 8 with 110 kg. Session 1: stable RPE 8. Session 2: same 110 kg, RPE 7.5. Session 3: even easier, RPE 7. Three sessions at stable or decreasing RPE on the same load mean one thing: that load no longer produces the target stimulus. Time to go up — on a compound at 110 kg, a sensible jump is 2.5 kg. Session 4: 112.5 kg, and you're off again.
This is pure autoregulation: the next load is driven by logged RPE feedback, not by a rigid percentage calculated on a 1RM tested six months ago.
Fatigue Management via RPE
The other side: RPE exposes fatigue. Same load, but today 110 kg comes out at RPE 9.5 instead of RPE 8. What happened? Poor sleep, aggressive caloric deficit, volume accumulated in previous sessions.
- Isolated drop (1 session): acute fatigue. Hold the load, cut a set.
- Persistent drop (2-3 sessions): systemic fatigue. Time for a deload or to cut volume.
RPE becomes the fatigue thermometer. Without it you only look at kg, and kg alone lie.
The Link With Progressive Overload
Progressive overload isn't just "add weight every week": it's the process of increasing the stimulus over time, and RPE is one of the cleanest tools to figure out when and how much to increase. Without an effort yardstick, you skip jumps you could make or you force them when you should hold steady.
The Practical Point of Autoregulation
Reviewing each progression against RPE data from previous sessions is simple in theory, annoying in practice: it means keeping a tidy log, comparing it, weighing trends. IRON uses the RPE you enter to suggest the load for your next session via the Overload Advisor — that's the practical point of autoregulation, automated on top of the data you already track. The principle stays yours: RPE is the key, the tool is just the amplifier.
9. 5 Mistakes That Ruin Your Use of RPE
1. Confusing It With Eccentric Difficulty/ROM
"Deep squats always feel like RPE 9 to me, even with three in the tank". You're confusing technical difficulty with effort. RPE measures how many reps you can still do, not how uncomfortable the movement is. A half squat with very heavy load can be RPE 8; a full ROM squat with moderate load, RPE 6.
2. Confusing It With DOMS
"My quads are wrecked from last week, today even the warm-up is RPE 8". DOMS isn't effort. Residual fatigue affects actual performance, but RPE has to be estimated on today's performance, not perceived discomfort. If at 80 kg you grind out 5 slow reps with the bar zigzagging on the last two, you're at RPE 9. Period.
3. Using It as a Pure Beginner in the First 6 Months
Unpopular but true: in the first six months of serious training RPE is a premature tool. Without an internal failure reference you misjudge it systematically. In that phase you need a simple linear progression (add 2.5 kg when you complete X reps) more than an autoregulation system. RPE comes into play once you've built the foundations and you're approaching your real limits.
4. Over/Under-Estimating Out of Ego or Anxiety
Two mirror biases: ego stim ("it was RPE 8" on a set actually at 10, to avoid admitting you couldn't hold the plan) and anxiety stim ("RPE 9" on a set you should have hit fresh, to give yourself permission not to push). Antidote: film sets close to your max. Bar speed leaves no room for bias.
5. Tracking by Feel Instead of by Method
Writing "set 1: heavy" isn't tracking RPE. Tracking it means putting down a number, every time, set by set. Without a written number you can't compare today's session with one from three weeks ago, and the whole autoregulation argument collapses. If you don't log it, it doesn't exist.
10. FAQ
Should I Use RPE as a Beginner?
No, not as a primary tool. In the first 4-6 months you don't have internal experience of failure to estimate RIR. Better to run a linear progression (+2.5 kg when you complete X reps with good form). You can familiarise yourself with the scale as a secondary metric — write the RPE post-set, but don't base load decisions on it. After 6-12 months of practice, RPE becomes genuinely usable.
RPE vs % of 1RM: Which Is Better?
False dichotomy, they're complementary. % of 1RM is objective but rigid — your "tested 1RM" is valid for that day, and a month later it can already be different. RPE is subjective but adaptive, it handles daily variations in form, recovery, sleep. Best practice: use the % for the rough plan ("70%" as starting load), use RPE as a real-time corrector (go up or down relative to the plan). It's exactly the approach of modern evidence-based templates.
How Do I Improve RPE Calibration?
Three actions in order of impact: film sets at RPE 8+ (reviewing real bar speed is the most effective tool), failure test every 4-8 weeks (take a set to true technical failure, measure the gap against your prediction), written log (RPE post-set always, only written data reveals patterns like "I overestimate the first set, underestimate the last"). Expect 2-3 months of deliberate practice for estimates with average error under 1 point.
Can I Use RPE on Every Exercise?
Technically yes, but with varying reliability. On compounds (squat, bench, deadlift, military press, row) it works well and is the main load-selection variable. On isolations (curl, lateral raises, leg curl) reliability drops: better rep targets to technical failure. On bodyweight there's RIR overestimation, use the speed of the last rep as a corrector. On cardio don't use 1-10, use Borg 6-20.
RPE and Cardio: Same Scale?
No, they're two different scales. The 1-10 with RIR (Tuchscherer, Zourdos) is anchored to muscular failure in lifting, and makes no sense on running or cycling, where "failure" is cardio-respiratory. For cardio use the Borg 6-20 scale: Borg 12-14 moderate aerobic zone, 15-17 threshold/intervals, 18-20 max effort/sprint. Keeping the two scales separate avoids the classic "RPE 8 over 5 km", which means nothing.
