How Many Sets Build Muscle? The Dose-Response Evidence

Finding one: the curve flattens, it does not fall

Hypertrophy rises with weekly volume across the studied range. The returns diminish — the tenth set does less than the fifth — but they diminish slowly, and far more slowly for size than for strength.

That asymmetry matters. Strength adaptations plateau in response to added volume much earlier than size adaptations do, which is why a powerlifter and a bodybuilder training for different outcomes are correct to run different volumes even when the movements are the same.

There is no cliff in the data at ten sets, or at twenty. What there is, is a curve that is still climbing at twenty and climbing slowly enough that the added fatigue and time cost start to dominate.

The practical range for a trained lifter: 12 to 20 fractional sets per muscle per week, starting at 12 to 14 and ramping. Below about 10 you are leaving growth on the table.

Finding two: frequency is volume-neutral

Training a muscle twice a week versus four times does not matter per se, once weekly volume is equated.

It matters indirectly, and the indirect mechanism is the whole practical value of the finding: frequency changes how much quality volume fits in a week. Twenty-five hard sets for one muscle in a single session are not twenty-five hard sets — the last ten are performed under accumulated local fatigue and are not the same stimulus as the first ten.

So the rule that follows is not "train each muscle N times a week". It is: pick the split that lets you hit your set target at 1 to 3 RIR without marathon sessions. For most people that lands at two sessions per muscle per week, run as an upper/lower four-day or a push/pull/legs six-day.

For a hybrid athlete it lands on upper/lower four days for a further reason: it leaves three days clear for endurance work. Choosing a split.

Finding three: fractional set counting

The best-fitting model was direct sets plus 0.5 × indirect sets.

A set counts in full for the muscle acting as prime mover and at half for the muscles assisting:

Two consequences people find surprising when they first count honestly.

Back volume is usually higher than assumed, because rows, pull-ups and deadlifts all contribute and the indirect contributions compound.

Biceps and triceps rarely need much direct work, because a normal pushing and pulling programme already delivers substantial indirect volume. Adding twelve direct sets of curls to a programme with fourteen sets of rowing puts you well past twenty.

The calculator does the arithmetic.

What counts as a set

Only sets taken close enough to failure to be a stimulus.

The proximity-to-failure relationship is non-linear: within roughly 0 to 5 RIR closer to failure is modestly better, and above about 5 RIR hypertrophy falls off steeply. A set at 6 RIR is not a low-quality working set. It is a warm-up, and counting it inflates your volume figure while contributing nothing.

This is where volume counting most often goes wrong in practice — not in the arithmetic, but in counting sets that were never hard enough to be sets. The proximity-to-failure evidence.

The "junk volume" confusion

The term is usually used to mean "volume past which added sets are worthless". The data does not show that within the studied range.

What it shows is diminishing returns, which is a different claim. At some point the next set returns so little hypertrophy that its cost in fatigue, time and recovery outweighs the return. That point depends on your training age, your recovery capacity, and — critically — on what else you are doing with that recovery capacity.

There is no universal number, and anyone quoting one as a hard ceiling is over-reading the evidence in the opposite direction from the people quoting ten sets as a floor.

What changes for a hybrid athlete

Everything above was measured in subjects whose only training stress was lifting. The recoverable ceiling is lower when a second stress shares the budget, and that ceiling — not the theoretically optimal volume — is what actually binds.

The failure mode is adding endurance work while holding lifting volume at the pure-lifter number. Twelve to fifteen sets you recover from and progress beat twenty you cannot repeat, and the dose-response curve is flat enough in that region that the difference costs you very little.

Limits of the evidence

Being straight about this is part of reading it correctly.

Most studies are 8 to 12 weeks. The question you actually have is about two years, and nobody has run that trial.

Untrained and trained subjects respond differently, and pooled analyses blur that — everything works in untrained subjects.

Measurement methods vary. Ultrasound muscle thickness at one site is not whole-muscle volume, and different studies measure different sites.

Volume was rarely counted fractionally in the primary studies, which is part of why the meta-regression's fractional model is a finding rather than a convention.

What to do next

Count one muscle honestly for a week, including the indirect work, and only counting sets that were within four reps of failure. Then compare against the band for your endurance load. Most people are either well below the range and calling it optimised, or above it and calling the resulting stall a plateau.