Both grow about the same amount of muscle, as long as the sets are taken close to failure. Across 28 studies of 747 healthy adults, light, moderate and heavy loads produced similar growth, while heavy loads produced more strength4.
The catch is effort. A light load only works when you push the set hard. Below: how these trials define light and heavy, what they found, and where the evidence stops.
What was studied
Load means how heavy the resistance is. Researchers describe it in two ways. One is RM (repetition maximum): an "8–12RM" load is one you can lift 8 to 12 times and no more. The other is a percentage of 1RM, the heaviest load you can lift once. In these studies, light usually meant around 20 to 35 reps per set, or 30% to 60% of 1RM; heavy meant around 8 to 12 reps, or more than 60% of 1RM.
Failure means the point in a set where another rep cannot be completed. Most load studies take every set to failure on purpose. That makes the comparison fair: both groups work to the same end point, only the load differs.
Growth is measured as muscle thickness or cross-sectional area on ultrasound or MRI scans, or muscle fibre size from a small tissue sample (biopsy).
| Study | Who | Light vs heavy | Length |
|---|---|---|---|
| Schoenfeld 2015 | 18 trained young men | 25–35 vs 8–12 reps, 7 exercises | 8 weeks |
| Morton 2016 | 49 trained young men | 20–25 reps (~30–50% 1RM) vs 8–12 reps (~75–90%) | 12 weeks |
| Lasevicius 2022 | 25 untrained men, one leg per condition | 30% vs 80% 1RM, to failure or not | 8 weeks |
| Hammert 2026 | 201 beginners, average age 21 | 20–30RM vs 8–12RM, single-arm curls | 6 weeks |
In plain terms: these are head-to-head trials where the only big difference is how heavy the load was.
What the studies found
Growth: a tie across the evidence. A 2017 meta-analysis of 21 studies, all with sets to failure and lasting at least six weeks, found hypertrophy was similar between loads at or below 60% of 1RM and loads above it3. A larger 2021 network meta-analysis pooled 28 studies of 747 healthy adults and split load three ways: light (more than 15RM), moderate (9 to 15RM) and heavy (8RM or fewer)4. No pair differed for growth. The standardised differences ranged from −0.09 to 0.15, which the authors called unlikely to be meaningful.
Trained lifters too. In 18 well-trained men, thickness gains were close between groups: 8.6% (light) vs 5.3% (heavy) for the elbow flexors, 5.2% vs 6.0% for the triceps, and 9.5% vs 9.3% for the quadriceps, with no significant differences1. In 49 trained men over 12 weeks of whole-body training, lean mass and fibre size rose with no significant difference between light and heavy groups2.
Light vs heavy loads in trained men: growth and strength gains · %
Light (25–35 reps) Heavy (8–12 reps)
Strength: heavy wins, mostly. Most of the trials that tied on growth split on strength. In the 2015 trial, back squat 1RM rose 19.6% with heavy loads and 8.8% with light ones1. The 2016 trial was the exception: 1RM rose in both groups, and only the bench press gain differed (14 kg heavy vs 9 kg light); its authors concluded load did not dictate strength gains "for the most part"2. In the 2021 pooled analysis, heavy loads beat light loads for 1RM strength by a standardised difference of 0.60 (95% CI 0.38 to 0.82), and moderate beat light by 0.34 (0.05 to 0.62)4. One likely reason, not tested in these trials: a 1RM test is itself a heavy lift, so heavy training is closer practice for it.
In plain terms: pick any load from light to heavy and growth looks similar. Maximal strength follows the load.
The condition: light sets have to be hard
Light loads matched heavy ones in studies where every set went to failure. What happens when they don't?
A 2022 trial answered that on one leg at a time. Twenty-five untrained men did knee extensions for eight weeks in four conditions: light (30% 1RM) or heavy (80%), each either to failure or stopped short, with total volume equated5. Thigh cross-sectional area grew 8.1% for heavy to failure, 7.7% for heavy not to failure and 7.8% for light to failure. Light loads stopped short grew only 2.8%, a change that was not significant.
Thigh muscle growth by load and effort, 8 weeks · % change in cross-sectional area
The light-load sets stopped short averaged about 20 reps, against about 34 when taken to failure5. Stopping at 20 with a load that allowed about 34 left many reps unused. Heavy sets stopped short, about 7 reps against about 12 to failure, grew as much as heavy sets to failure. This fits the wider evidence that ending sets near failure is what matters, not the final rep (see do you have to train to failure).
In plain terms: with a light load, "close to failure" means close to the real limit, which may be 30 reps away, not where it first starts to burn.
Could light loads even have the edge?
The newest trial, and the largest, leaned towards light. In 2026, 201 beginners averaging 21 years were randomised to four groups6. Two did four sets of single-arm curls to failure, three times a week for six weeks: one at 20–30RM (light), one at 8–12RM (heavy). A third group only did the tests plus a 1RM test halfway; a fourth did the tests only.
Elbow-flexor thickness rose about 0.30 cm with light loads and about 0.20 cm with heavy loads, and both beat the two non-training groups6. The authors wrote that growth "may be augmented" by light-load failure training. They also scanned the biceps tendon: it thickened slightly (about 0.02 to 0.03 cm) at one site in all three groups that lifted anything, including the halfway-test group, with no detectable change at the other site.
This is one trial, in beginners, on one arm exercise, for six weeks. It adds weight to "light works"; it does not show that light beats heavy in general. For picking a starting load in practice, see how much weight to start with.
The studies at a glance
| Study | Who was studied | Key result | Year |
|---|---|---|---|
| Schoenfeld et al., *J Strength Cond Res*1 | 18 well-trained young men, 8 weeks | Similar thickness gains; squat 1RM +19.6% heavy vs +8.8% light | 2015 |
| Morton et al., *J Appl Physiol*2 | 49 trained young men, 12 weeks | Lean mass and fibre size gains did not differ by load; of the 1RM tests, only bench press favoured heavy | 2016 |
| Schoenfeld et al., *J Strength Cond Res*3 | 21 studies, sets to failure, 6+ weeks | Hypertrophy similar; 1RM gains greater with heavy loads | 2017 |
| Lopez et al., *Med Sci Sports Exerc*4 | 28 studies, 747 healthy adults | No growth difference between loads; strength favoured heavy (0.60) | 2021 |
| Lasevicius et al., *J Strength Cond Res*5 | 25 untrained men, 4 leg conditions, 8 weeks | Light loads grew only when taken to failure (2.8% vs about 8%) | 2022 |
| Hammert et al., *Med Sci Sports Exerc*6 | 201 beginners, curls, 6 weeks | Light about 0.30 cm vs heavy about 0.20 cm thickness gain | 2026 |
Limits worth knowing
- Short trials. Six to twelve weeks. Whether light and heavy stay level over years is not tested.
- Mostly young men. Several trials enrolled only men; the 2026 trial did not report the sex split in its abstract.
- Small groups. The trained-lifter trial had nine people per group1; small trials can miss small differences.
- Failure in the lab. Light sets to failure are long and uncomfortable. The trials were supervised; whether people push light sets that hard on their own was not measured.
- Abstract only. We read the 2026 trial, both Schoenfeld papers, the 2016 trial and the 2022 trial as abstracts; the 2026 trial's per-group means, intervals and strength results are not in its abstract and are not stated here. The 2021 network meta-analysis was read in full.
Latest research
The newest trial is the 2026 curl study in 201 beginners, published online in June 20266. It is the largest head-to-head load trial we found, it had two non-training comparison groups, and it found slightly more growth with light loads. Its authors frame this carefully ("may be augmented"). Because it only tested one exercise in people new to lifting, it refines the long-standing tie rather than overturning it.
The short version
- Light and heavy loads grow similar muscle when sets are taken to or near failure.
- Heavy loads build more maximal strength.
- Light loads stopped far from failure grew much less in a direct test.
- One large 2026 trial in beginners found slightly more growth with light loads.
Educational overview only — not medical advice.
