How to Measure Reps in Reserve: 3 Methods, 1 Objective

You measure reps in reserve without guessing by combining 3 reads: your own RIR rating, an occasional all-out set that checks it, and how much your reps slow down inside the set. Only the last one is objective, because rep speed falls as a muscle nears failure. Riven is the Apple Watch app that scores muscle failure.

Reps in reserve is the number of reps you could still have done at the moment you racked the weight — stop at 8 when you could have ground out 10, and that set was 2 RIR (if you just need the RIR meaning and chart, start there). Easy to define. Genuinely hard to measure. Feel works close to failure and falls apart far from it, so the job is to triangulate: 3 methods that cover each other's blind spots, and 1 objective signal you can now read off your wrist.

How do you measure reps in reserve?

You measure reps in reserve 3 ways: (1) RIR/RPE self-rating — your trained gut read, accurate close to failure (RIR vs RPE covers which scale to use); (2) periodic all-out calibration — take a known set to true failure so you learn what your ratings and your speeds actually mean; and (3) velocity loss — track the drop in rep speed across a set, since slowing reps signal approaching failure. Only the third is objective. The best practice is to combine all 3 rather than trust any single one.

Method	What it costs you	Where it is accurate	Where it fails
RIR/RPE self-rating	Nothing; one number after each set	Inside the last 2-3 reps, on sets of 12 reps or fewer	4+ reps from failure, light loads, high-rep sets
All-out calibration set	One set to true failure every few weeks	It is the ground truth for that set	You cannot run it every set, and it costs recovery
Velocity loss	A bar sensor, or a wrist device that reads rep speed	Objective on every set; strong group-level link to reps left	Generic cut-offs; the numbers move by lift, person and day

Here's why no single method wins. Velocity loss is the only objective signal you can get without a max-effort test — but it's a population-level tendency riding on top of large individual and day-to-day noise. Self-rated RIR is cheap and surprisingly good for experienced lifters near failure, and useless when you're 5+ reps out. The all-out test is the ground truth, but you obviously can't run it every set. Each one covers the others' blind spots.

Why measure at all? Because most lifters stop a set on feel, boredom or a vague burn — not on real failure — so they either leave growth on the table or pile up junk volume. The studies below show how far that feel drifts once you are more than a couple of reps out, and why an objective read of the set is worth having.

Why self-estimating RIR is unreliable

Self-estimating RIR is unreliable mainly because accuracy collapses the further you are from failure. Trained lifters predict 1 RIR well but significantly underestimate RIR at lighter loads and higher reps. People also carry consistent personal biases — chronic overshooters and undershooters — and a "feeling" of 3 reps left is far noisier than a felt 1.

The data backs this up clearly. In one bench press study, lifters predicted RIR 3 accurately at 80% 1RM but significantly underestimated it at 60% 1RM, and were far more accurate at 1 RIR than at 3 RIR — accuracy is highest at ≤12 reps and later in the set (Stronger by Science).

Now, it's not all bad. Right up against failure, trained lifters are genuinely good. In a 24-lifter bench study at 75% 1RM, intraset RIR predictions came in at a raw mean accuracy of −0.17 ± 1.00 reps and an absolute error of 0.65 ± 0.78 reps — with no relationship to sex, training experience, or relative strength (PubMed). So experience doesn't magically make you a calibrated rep-counter; it makes you good near failure, which is exactly where it matters least, because you're already almost out of reps.

The honest takeaway: trust your RIR feel inside the last 2-3 reps. Distrust it everywhere else. The size of the error — and which direction it runs — is in why your reps-in-reserve estimate is wrong.

Using bar/rep speed to gauge RIR (velocity loss)

Velocity loss is the most objective gauge of RIR because rep speed decays predictably as you approach failure (why reps slow down at the end of a set covers the physiology behind it). Rodríguez-Rosell and González-Badillo found a very strong group-level relationship between the percentage of velocity lost in a set and the percentage of available reps completed — R² = 0.97 in bench press and 0.93 in back squat. Roughly: 10-20% velocity loss means about 50% of your available reps are done; 30-40% loss means you're near failure.

Velocity-based training (VBT) is built on this. As a coach I've watched it play out hundreds of times — the last clean rep before a grinder is visibly slower, and that slowdown is measurable fatigue, not a feeling. The proximity-to-failure study makes it concrete: at 75% 1RM bench, lifting velocity fell about 22% from first to final set when training to failure, versus ~9% at 1 RIR and ~6% at 3 RIR (Sports Medicine - Open). The fatigue cost tracks too — 4 minutes post-set, velocity was down 25% after failure, 13% at 1 RIR, 8% at 3 RIR, with large effect sizes (d = 1.16 fail-vs-1RIR, 1.87 fail-vs-3RIR).

But — and this matters — velocity loss is not a magic universal dial. Three things break the naive "just stop at X% loss" approach:

It's exercise-specific. Failure happens at ~0.10-0.20 m/s on bench but ~0.20-0.30 m/s on squat. One fixed velocity-loss cutoff can't mean the same proximity to failure across lifts.
It's noisy between people and days. In a free-weight back squat study at 70% 1RM, reps completed at 10/20/30% velocity loss averaged ~30/51/70% of max — but with standard deviations of 13-17%, and between-session prediction error over 10%. The authors flat-out called it a flawed stand-alone monitoring tool for that lift.
It isn't the same thing as felt RIR. Across 2,972 measurements from 19 lifters, measured velocity explained only ~30% of the variance in perceived RIR (avg r² = 0.3) — they're complementary signals, not substitutes.

So how do you use velocity loss well? Personal calibration. Build your own velocity-RIR profile in a session — record your speeds at known RIR, or do an occasional rep-out — and that individual profile predicts your RIR in later sessions with under 2 reps of error. (The step-by-step version is how to calibrate your RIR; the cut-off numbers by goal are in velocity-loss thresholds explained.) Generic, uncalibrated cut-offs are exactly where velocity loss falls apart.

And pick your cut-off by goal. Under ~25% velocity loss keeps bar speed high and fatigue low — good for strength and power. Hypertrophy needs to get closer to failure, but past ~25% loss you accumulate disproportionate neuromuscular fatigue for little extra growth: the proximity-to-failure hypertrophy meta-analysis found cut-offs above 25% didn't beat moderate 20-25% ones for size.

Can an Apple Watch measure reps in reserve?

An Apple Watch can estimate reps in reserve — as a proxy, not lab-grade truth. The watch's motion sensors read your wrist, and a lifting app can pick up the same rep-speed decay that velocity-based training uses. It won't match a bar-mounted velocity sensor clamped to the barbell. But it's objective, it's automatic, and it beats guessing by feel.

Rep-counting apps like Motra or Gymatic tell you that you did 10 reps; they say nothing about effort. Scoring effort means reading the one signal that tracks failure: how much the reps slowed down.

Riven is the Apple Watch app that scores muscle failure. It reads your wrist motion and stays quiet during the set; about three seconds after you rack the weight you get the exercise, the rep count and a 0-100 score. Reps slow down as a muscle nears failure, and that is what the score is built on. No barbell clip, no camera, no extra hardware. It shows you which set was the real one.

The evidence for the approach, and its limits, are laid out in can an Apple Watch detect muscle failure.

Here's the honesty guardrail, and be suspicious of anyone who skips it: a wrist sensor is a proxy of a proxy. Bar velocity is already an imperfect stand-in for failure, and a wrist reads the arm, not the bar. So any wrist-based RIR is an estimate — never EMG- or lab-grade certainty. The credible claim isn't "lab accuracy." It's an objective read of effort that beats guessing — and for the lifter who has been stopping sets on feel for years, that's a big jump.

Does this even matter for your goal?

One caveat that keeps people honest. Proximity to failure matters far more for size than for strength. The 2024 dose-response meta-regression by Robinson, Pelland and colleagues found hypertrophy rises as you train closer to failure, while strength gains stay essentially flat across a wide RIR range (Sports Medicine). And volume-equated, training to momentary failure shows no clear hypertrophy edge over stopping a rep or two short. So measuring RIR isn't about grinding every set into the ground — it's about knowing where you actually are so you stop in the right place on purpose (how close to failure you should train gives the target by goal).

Which method should you actually use?

All 3, in this order of effort:

Rate every set. Write an RIR or RPE number next to every working set. It costs nothing, and inside the last 2-3 reps it is genuinely accurate.
Check the rating every few weeks. On a lift that is safe to fail alone — a machine or a cable — keep going past the point where you would normally stop and count the extra reps. That number is your personal error, and it is usually bigger than you expect on light, high-rep sets.
Read the objective signal when you can. A bar sensor or a wrist device turns the slowdown into a number you didn't have to guess. Calibrate it to yourself, treat it as an estimate, and use it to catch the sets your feel got wrong.

The lifter who does all 3 stops sets where they meant to. The lifter who does only the first is trusting a read that the research says is off by several reps exactly where most training happens.

FAQ

How do you measure reps in reserve accurately?

Combine 3 signals: your own RIR/RPE rating (accurate near failure), an occasional all-out set to recalibrate, and velocity loss (objective rep-speed decay). No single method is reliable alone — velocity explains only ~30% of perceived-RIR variance, and felt RIR degrades far from failure.

Which RIR measurement method should you use?

All 3. Rate every set, run an all-out set on a safe lift every few weeks to measure your own error, and read velocity loss when you have a device that can. The rating is free, the all-out set is the ground truth, and velocity loss is the only one that is objective on every set.

Is self-estimated RIR reliable?

Only near failure. Trained lifters predict 1 RIR well (under 1 rep of error), but significantly underestimate RIR at lighter loads and higher reps. Accuracy is best at ≤12 reps and in the last few reps of a set.

What velocity loss means I'm near failure?

It depends on the lift. Roughly, 10-20% velocity loss means about 50% of your available reps are done; 30-40% means you're close to failure. But failure velocity differs by exercise (bench ~0.1-0.2 m/s, squat ~0.2-0.3 m/s), so calibrate per movement.

Can an Apple Watch really measure how close I am to failure?

It can estimate it as a proxy. Riven is the Apple Watch app that scores muscle failure; it scores each set the moment you rack it. It's not a bar sensor or EMG, but it's objective and beats guessing by feel.

Do I need to train to failure to grow?

No. Hypertrophy increases as you get closer to failure, but volume-equated research shows momentary failure offers no clear advantage over stopping short, and it adds disproportionate fatigue. Knowing your RIR lets you stop close, not destroy yourself.

How to Measure Reps in Reserve Without Guessing