Add reps or add load?
In 38 trained young adults over 8 weeks of leg work to failure, no clear difference: squat 1RM 2.0 kg toward load (90% CI −2.4 to 7.8). No increment rule given; not tested in endurance athletes.
Main caveat: Short and specific. 8 weeks, lower body only; the authors exclude inference to other muscles, older people, adolescents, the untrained and elite lifters.
Plotkin and colleagues randomised 43 lifters with at least one year of lifting to two ways of progressing the same four lower-body exercises: four sets each, twice a week, for 8 weeks, starting from a load found by a 10RM test.
The LOAD group "aimed to increase load while maintaining this target repetition range" of 8-12RM; the REPS group "aimed to increase the number of repetitions performed per set while maintaining the initial load". Both were encouraged to go to momentary failure.
Of 38 finishers, the difference between groups was small on nearly everything: muscle thickness under 1 mm except rectus femoris, leg-extension reps about 2%, jump 0.1 cm, body fat under 1%. Rectus femoris thickness summed across three sites favoured REPS by 2.8 mm (90% CI −0.5 to 5.8), and Smith-machine squat 1RM favoured LOAD by 2.0 kg (90% CI −2.4 to 7.8).
Both 90% intervals cross zero, and the strength interval is wide relative to its estimate. The authors conclude both are "viable"; read that as no clear difference, not sameness. Our reading: an honest first trial and a real-world example of a simple rep-range scheme. It does not say how much to add or when, and the rep-progression arm's final reps are not reported.
Protocol
Design: parallel randomised trial, block randomisation; preregistered (osf.io/yvhcs). Sample set by a precision analysis on mid-thigh thickness, 18 per group wanted.
Who: n = 43 randomised (27 men, 16 women): LOAD n = 22 (13 men, nine women); REPS n = 21 (14 men, seven women). Age 23.1 ± 5.3 years, 77.2 ± 16.7 kg, 3.8 ± 4.0 years of training. Lifting at least three times a week on most weeks for at least a year, lower body at least weekly; squat experience not required, and many did not squat regularly.
Finishers: 38 (LOAD n = 21; REPS n = 17). Dropouts: personal reasons n = 2, lack of compliance n = 2, training-related injury n = 1 (group not stated in the text). Attendance LOAD 94.9%, REPS 95.2%.
Training: free-weight back squat, leg extension, straight-leg calf raise, seated calf raise; 4 sets each; 2 sessions a week on non-consecutive days; 8 weeks; 2 min rest; concentric about 1 s and eccentric about 2 s; supervised. A shared upper-body programme done unsupervised on other days.
Increment rule, exactly as given: a 10RM test set first loads. Session 1 for both aimed at 8-12RM per set. In later sessions, LOAD aimed to increase load while staying in the range; REPS aimed to increase reps per set at the initial load. All sets verbally encouraged to momentary concentric failure. No size of a load step, no trigger, and no cap on reps is stated.
Outcomes: ultrasound thickness of quadriceps at three sites (rectus femoris and vastus lateralis composites), gastrocnemii, soleus; Smith-machine squat 1RM; leg-extension reps at 60% of body mass; countermovement jump; bioimpedance body fat and leg lean mass. Isometric strength was planned but dropped for dynamometer calibration problems.
Analysis: regression on change, baseline and sex as covariates; Poisson for reps; 90% CIs by bootstrap with 5,000 replicates; no null-hypothesis tests; leave-one-out sensitivity.
Funding and interests: PSC CUNY grant from the State of New York; funder had no role. B. J. S. serves on the scientific advisory board of Tonal Corporation, a manufacturer of exercise equipment. M. I. and J. F. are employed by Renaissance Periodization.
Key Results
| Outcome | LOAD change | REPS change | Adjusted effect, REPS vs LOAD (90% CI) |
|---|---|---|---|
| Smith squat 1RM (kg) | 21.8 ± 21.2 | 19.3 ± 7.7 | −2.0 (−7.8 to 2.4) |
| Leg-extension reps at 60% body mass | 6.6 ± 3.0 | 6.8 ± 5.5 | 1.02 times (0.93-1.14) |
| Countermovement jump (cm) | −0.1 ± 2.7 | −0.1 ± 3.4 | 0.1 (−1.5 to 1.7) |
| Rectus femoris, sum of sites (mm) | 10.1 ± 7.9 | 12.3 ± 5.1 | 2.8 (−0.5 to 5.8) |
| Vastus lateralis, sum of sites (mm) | 10.9 ± 7.1 | 10.2 ± 6.1 | −0.3 (−3.6 to 3.6) |
| Lateral gastrocnemius (mm) | 1.4 ± 1.2 | 1.0 ± 0.8 | −0.2 (−0.8 to 0.3) |
| Medial gastrocnemius (mm) | 1.1 ± 1.3 | 1.5 ± 2.7 | 0.5 (−0.4 to 2.2) |
| Soleus (mm) | 1.1 ± 1.1 | 1.1 ± 1.3 | 0 (−0.6 to 0.7) |
| Leg lean mass (kg) | 0.3 ± 0.4 | 0.3 ± 0.3 | 0.1 (−0.1 to 0.3) |
| Body fat (%) | −0.5 ± 2.7 | −0.8 ± 2.3 | −0.4 (−1.7 to 1.1) |
Positive favours REPS. CIs are 90%, as the authors chose; a 95% interval would be wider.
Baseline: Smith squat 1RM 76.9 ± 29.9 kg (LOAD) and 86.8 ± 27.1 kg (REPS); the model adjusted for it.
Size of gains: pooled muscle-thickness gains of 6.7% to 12.9%; squat 1RM up about 20 kg, which the authors partly assign to initial neuromuscular gains, since many did not squat regularly.
Jump: did not improve in either group.
Sensitivity: no single participant shifted the conclusions.
Mechanisms
Measured: thickness and 1RM after 8 weeks; no neural or fibre measures.
Authors' interpretation: their hypothesis was that effort and working sets drive hypertrophy, so similar growth was expected, and they predicted load would win on maximum strength and reps on muscular endurance. They cite a retrospective, non-randomised analysis favouring load for vastus lateralis size (16.0 ± 4.0% vs 7.9 ± 4.0%).The rectus femoris lean is speculated to come from fatigue in the vasti at high reps, and could be chance. The missing strength gap may reflect specificity: training with free weights and testing on a Smith machine, both groups far from 1RM.
Inference: REPS kept the starting load, so its reps must rise over the block while LOAD's stay in the range; heavy-strength goals, and rep bands of 4-8 used for endurance athletes in the base, are not what this tested.
Practical Application
Add reps or add load? In 38 trained young adults over 8 weeks of leg work to failure, no clear difference: squat 1RM 2.0 kg toward load (90% CI −2.4 to 7.8). No increment rule given; not tested in endurance athletes.
Real-world example of a rule, as far as the paper goes: start at a load that gives about 8-12 reps, then either keep reps in the range and add kg when you can, or keep kg and push reps. It does not say how many kg or when.
When reps are the honest choice: the authors call rep progression "particularly useful when greater loads may not be available"; our examples (inference): bodyweight sets, a fixed dumbbell or a home gym, or a jump too big for the next plate. This trial gives some support that progression need not wait for added load, within its range and effort.
For a heavy-strength goal, stay cautious: the 1RM interval runs up to 7.8 kg in favour of load, so the data do not rule out a load advantage; REF-1476 also lists heavier loads for strength. This is a reason to keep load as the default when strength is the goal, while reps remain open.
Limitations
Short and specific. 8 weeks, lower body only; the authors exclude inference to other muscles, older people, adolescents, the untrained and elite lifters.
The failure instruction was not fully met. Some stopped short, and REPS "appeared to have greater difficulty approaching true failure". No RPE or reps-in-reserve data were collected.
Testing. Trained on free weights, tested on a Smith machine; strength testing not blinded; the squat wasn't familiar to many, so learning may colour the gains. LOAD's change SD in squat 1RM (21.2 kg) was much wider than REPS's (7.7 kg), in a small sample.
Attrition. 4 of 21 lost in REPS against 1 of 22 in LOAD (our arithmetic); the one training-related injury is not assigned to a group.
Progression not quantified. Neither the load added in LOAD nor the reps reached in REPS is reported in the text, so the actual dose of progression is unknown, as are the increments. Volume load was not reported.
Interests. One author on an equipment maker's advisory board; two employed by a training-programme company. Public university grant.
How this entry was checked and changed
Why this level
read from the full text (open access, CC BY). 43 randomised, 38 analysed, one lab; rubric's parallel RCT 20-39 by those who finished. Assessor and statistician blinded for ultrasound; strength tests not blinded
Full text (checked 2026-10-30)
n = 43 randomised: LOAD n = 22; REPS n = 21; 38 completed the study (LOAD n = 21; REPS n = 17); attendance 94.9% and 95.2% (full text)
Age 23.1 ± 5.3 years, training experience 3.8 ± 4.0 years; four sets of four exercises, two weekly sessions, 8 weeks, 2 min rest (full text)
Squat 1RM (kg): LOAD 21.8 ± 21.2, REPS 19.3 ± 7.7, adjusted effect −2.0 [−7.8 to 2.4]; CIs are 90% (full text)
RF sum (mm): 10.1 ± 7.9 vs 12.3 ± 5.1, adjusted effect 2.8 [−0.5 to 5.8]; VL sum: −0.3 [−3.6 to 3.6] (full text)
Endurance reps: 1.02 [0.93–1.14]; CMJ: 0.1 [−1.5 to1.7] (full text)
LOAD aimed to increase load while maintaining the target range; 8–12 repetition maximum; REP group aimed to increase repetitions at the initial load; no load increment stated (full text)
Pooled mean muscle-thickness increases ranged from 6.7% to 12.9% (full text)
Funded by PSC CUNY Award # 64346-00 52; Schoenfeld on Tonal Corporation's advisory board; Israetel and Feather employed by Renaissance Periodization (full text)