How muscle grows

• Sets: about 10 or more hard sets per muscle group per week; as few as 4 still works.
• Effort: finish most sets with 0 to 3 reps in reserve.
• Load: anything from about 5 to 30 reps builds muscle if the set is hard. Go heavy for strength.
• Frequency: train each muscle at least twice a week, mainly because it's practical.
• Rest: at least 60 to 90 seconds; 2 to 3 minutes on heavy compound lifts.
• Progression: add reps or load over time.

The ACSM's 2026 position stand, an overview of 137 systematic reviews covering more than 30,000 participants, reached a humbling conclusion: few programming variables consistently change outcomes. Doing progressive resistance training at all is the big win; the rules below are refinements ^[1].

Weekly set volume has the clearest dose-response relationship with growth. In a 2017 meta-analysis, each additional weekly set per muscle was associated with roughly 0.37 per cent more growth, and 10 or more weekly sets tended to outperform fewer than 5 ^[2]. The ACSM position stand lands on the same figure: hypertrophy is enhanced at 10 or more sets per week ^[1].

A 2026 meta-regression of 67 studies confirmed that more volume means more growth, but with diminishing returns. It also found results are best predicted when sets that train a muscle indirectly count as half a set, so three sets of rows count as about one and a half sets for biceps ^[3].

Practical range: 10 to 20 weekly sets per muscle, counted that way. Beginners can start lower. If time is short, a review of time-efficient training suggests a floor of about 4 weekly sets per muscle group ^[4].

Confidence: high that more sets beat fewer up to moderate volumes. Low on where the ceiling sits; very high volumes are less studied and recovery differs between people.

A set only counts if it's hard. Reps in reserve (RIR) is the number of good reps you had left when you stopped.

Meta-analyses comparing failure with non-failure training find no clear advantage to reaching failure for muscle growth ^{[5] [6]}. A 2024 meta-regression went further and estimated RIR across studies: growth increased as sets finished closer to failure, while strength gains were similar across a wide range of RIR. The authors flag the analysis as exploratory ^[7].

Practical range: 0 to 3 RIR for most sets. Stay at 1 to 3 on heavy squats, deadlifts and presses, where failure costs more fatigue and form; go to 0 or 1 on machines, cables and isolation lifts.

People misjudge RIR by about one rep on average, usually thinking they're closer to failure than they are, and accuracy improves near the end of a set and with lower rep counts ^[8]. Taking the last set of a safe exercise to true failure now and then is a useful calibration.

Confidence: moderate.

When sets are taken to failure, light loads (60 per cent of one-rep max or less) and heavy loads build similar muscle. Heavy loads build more maximal strength ^[9]. The ACSM stand agrees: strength is enhanced by loads of 80 per cent and above ^[1].

So for muscle, pick the rep range by practicality. Very high reps are uncomfortable and make it hard to judge effort; very low reps need many sets and load the joints. Most work fits well between 6 and 15 reps, with up to 30 or so acceptable on isolation lifts if you push close to failure ^[4].

Rep speed matters little: rep durations from 0.5 to 8 seconds produce similar growth, and deliberately very slow reps (over 10 seconds) look worse ^[10]. Lower under control, lift with intent.

Confidence: high.

With weekly sets held equal, training a muscle once, twice or three times a week produces similar growth ^[11]. The 2026 meta-regression agrees that frequency's effect on hypertrophy is compatible with negligible, though strength does improve with more frequent practice ^[3].

Frequency matters indirectly. Twelve hard sets for one muscle in a single session is a long, grinding session and the late sets suffer. Splitting it over two or three days keeps set quality high.

Practical range: each muscle 2 times a week; more if your volume is high.

Confidence: high for the volume-equated finding.

Short rests reduce the reps you can do on later sets. A 2024 Bayesian meta-analysis found a small hypertrophy benefit to resting more than 60 seconds and no detectable extra benefit beyond 90 seconds ^[12]. For maximal strength in trained lifters, rests longer than 2 minutes appear necessary ^[13].

Practical range: 1 to 2 minutes for isolation and machine work, 2 to 3 minutes or more for heavy compound lifts. If you're short of time, superset opposing muscles in place of cutting rests ^[4].

Confidence: moderate. Only nine studies met the criteria for the hypertrophy analysis.

Muscle adapts to what it's asked to do, then stops. Progressive overload means the work gets harder over time: more reps with the same weight, more weight for the same reps, or more sets.

Adding reps and adding load worked about equally well over eight weeks in a head-to-head trial ^[14]. The ACSM's older rule of thumb still holds: when you can do one or two reps over your target, increase load by 2 to 10 per cent ^[15].

Confidence: high on the principle; the specific method is a matter of preference.

Two inputs outside training have good evidence.

Protein. In a meta-analysis of randomised trials, extra protein increased gains in muscle and strength, with no further benefit beyond a total intake of about 1.6 g per kg of body weight per day ^[16].

Sleep. Short sleep impairs maximal strength in compound lifts ^[17] and physical performance generally ^[18]. Its direct effect on long-term muscle growth is less studied.

This is general information. If you have a medical condition, an injury, or you're unsure how to start, see your GP or a qualified exercise professional first.