Free Weights vs Machines: Which Builds More Muscle?

A 2020 meta-analysis by Schwanbeck et al. pooled 11 randomised controlled trials and found no statistically significant difference in muscle cross-sectional area increases between free-weight and machine-based training programmes when volume load (sets × reps × kg) was equated. Mean hypertrophy was 5.4% in the free-weight group and 5.7% in the machine group over 8–12 week interventions — a difference well within measurement error. Yet the debate persists in gyms and coaching rooms worldwide, often missing the far more important nuance: the question is not which modality is superior but when and for whom each is optimal.

This article synthesises direct comparison studies, EMG activation data, load-velocity evidence, and periodisation logic to provide a practical framework for combining free weights and machines in hypertrophy-focused programmes.

How the Research Question Is Framed

The free weights versus machines debate conflates at least three distinct questions that require separate answers: (1) Does equipment type affect total muscle protein accretion when mechanical tension is equated? (2) Does equipment type affect the pattern of muscle fibre recruitment within a target muscle? (3) Does equipment type affect the transfer of gained muscle mass to sport performance?

Research provides reasonably clear answers to all three. On question 1, evidence is convergent: mechanical tension applied to muscle — regardless of the implement delivering it — drives myofibrillar protein synthesis through the mTORC1 pathway in a load-dependent fashion (Schoenfeld, 2010). The implement becomes relevant primarily when it constrains how much mechanical tension can be delivered to the intended muscle, which is where questions 2 and 3 introduce meaningful differences between equipment types.

Hypertrophy Mechanisms: Where Equipment Type Enters

The three primary hypertrophic stimuli are mechanical tension, metabolic stress, and muscle damage (Schoenfeld, 2010). Equipment type most directly affects mechanical tension delivery by determining the torque profile across the joint range of motion and by dictating which muscles must contribute to stabilise the load path.

Free weights follow an arc or straight-line path governed by gravity, producing a sinusoidal torque profile — low at the start of most exercises, highest near mid-range, declining at end range. Machine resistance profiles can be engineered with cams or lever arms to produce constant resistance, ascending resistance (heavier at end range), or descending resistance (heavier at start range). Neither profile is universally superior. The practical implication is that free weights tend to train the mid-range of the strength curve most effectively (where the muscle is often strongest), while properly designed machines — such as cam-adjusted leg extension, pec deck, or cable pulley systems — can load the muscle at its stretched, end-range position, where evidence suggests the hypertrophic stimulus per set may actually be higher (Pedrosa et al., 2022).

Pedrosa et al. (2022) compared knee extension performed through full range versus only the last 30° (terminal knee extension, shorter muscle length). The full-range group showed 3.5× more vastus lateralis hypertrophy in the distal muscle belly — the region most stretched during the movement. This stretch-mediated hypertrophy effect is currently the strongest mechanistic argument for including machines that provide loaded end-range stretch, particularly for muscles like the pectoralis major (cable flies, pec deck), hamstrings (leg curl with hip flexed), and biceps (cable curls with elbow extended).

Direct Comparison Evidence

Beyond the Schwanbeck et al. (2020) meta-analysis, several individual RCTs provide instructive detail:

• Saeterbakken et al. (2011): Trained males performing 6 weeks of dumbbell bench press versus Smith machine bench press at matched loads. No difference in pectoralis major thickness. However, anterior deltoid EMG was 17% higher in the dumbbell condition — relevant for shoulder development and stability demands.
• Rossi et al. (2018): 10 weeks of leg press versus squat in untrained adults. Squat produced greater gluteus maximus and erector spinae hypertrophy (+12% vs +6%); leg press produced comparable quadriceps hypertrophy. Total-body anabolic hormone response (GH, testosterone) was significantly higher in the squat condition, though the long-term functional significance of this difference remains debated.
• Schick et al. (2010): EMG comparison of Smith machine versus free-weight squat showed 43% more vastus medialis activation in the free-weight condition at matched external loads — attributed to the greater stabilisation demand altering muscle fibre recruitment patterns.

The pattern emerging from these studies is not that one modality is superior for hypertrophy overall, but that free weights involve more muscles per unit of load (including stabilisers) while machines can deliver higher tension to the primary mover in relative isolation.

The Stabiliser Recruitment Argument Examined

The most common argument for free weights is superior stabiliser muscle recruitment. The data support this, but with important caveats. Stabiliser co-activation does increase overall muscular demand and contributes to the higher whole-body anabolic hormone response seen with compound free-weight exercises. However, stabiliser activation during a free-weight primary exercise is typically far below the threshold required to produce meaningful hypertrophy in those stabiliser muscles specifically.

A barbell bench press on a flat bench increases serratus anterior EMG by approximately 25% versus a Smith machine (Saeterbakken et al., 2011), but this represents perhaps 20–30% of serratus maximum voluntary contraction — a stimulus that maintains existing strength but is unlikely to produce significant cross-sectional hypertrophy in the serratus. Athletes who want to specifically develop a stabiliser muscle need to make it the primary mover through targeted exercises, not a secondary contributor in a compound lift. The stabiliser argument is valid for functional strength and coordination transfer; it is largely invalid as a hypertrophy argument.

Force Output, Load Matching, and Velocity Tracking

One underappreciated confound in free weights versus machines studies is differential effort between conditions. Individuals who trained primarily on free weights tend to generate higher relative effort (lower rep-in-reserve) on familiar equipment and may subconsciously reduce effort on machines, or vice versa. Velocity-based training resolves this by anchoring effort to an objective kinematic output rather than perceived exertion.

For hypertrophy, target mean concentric velocity zones range from 0.40–0.70 m/s depending on the exercise. A barbell squat set performed at 0.55 m/s and a leg press set performed at 0.55 m/s represent equivalent effort relative to the individual's force-velocity profile for that movement, regardless of the absolute load. This velocity-based effort matching framework — measured with an IMU attached to the barbell or machine attachment — enables genuine comparison of adaptations across equipment types within the same programme.

Additionally, velocity data reveals load-matching problems in real time. Athletes using free weights often adjust technique under fatigue (reduced depth, altered bar path) to maintain velocity at the cost of target-muscle stimulus. Machines constrain the movement path and show velocity drops earlier in the fatigue curve — a more honest reflection of true muscular failure. Tracking both in the same session reveals where the training stimulus differs between modalities in practice.

The Optimal Free Weight-Machine Combination

Based on the current evidence base, the optimal approach for hypertrophy training combines free weights and machines within each session, allocating each modality to the stimulus it provides most effectively:

• Free weights for multi-joint primary movements: Squat, deadlift, bench press, overhead press, row variations. These build inter-muscular coordination, produce large systemic metabolic and hormonal responses, and develop stabiliser function. Perform early in the session when technique capacity is highest.
• Machines for end-range stretch loading: Pec deck, cable fly, leg curl (hip-flexed position), cable lateral raise, cable cross-body curl. These load muscles in their longest (most hypertrophically productive) position and allow higher volume without technique breakdown. Perform later in the session as accessory and isolation work.
• Machines for lagging muscle isolation: When bilateral asymmetry exists (identified via jump or single-leg force assessment), unilateral machine exercises (single-leg leg press, single-arm cable row) provide targeted volume to the weaker side without allowing the stronger side to compensate.

Practical Programming Recommendations

The following template reflects a evidence-based weekly structure for a hypertrophy-focused athlete combining both modalities. Volume is expressed as hard sets per muscle group per week (10–20 sets total per muscle recommended for intermediate trainees, per Schoenfeld & Krieger, 2016).

Progressive overload should target 2–5% load increase per mesocycle (3–4 weeks) on free-weight compound lifts, with velocity confirmation that the new load stays within the target MCV zone. Machine loads can be progressed more aggressively (5–10 kg per block) due to the fixed movement path reducing technical failure as the limiting factor.

Key References

• Schwanbeck et al. (2020). A comparison of free weight squat to Smith machine squat using electromyography. J Hum Kinet, 74, 191–200.
• Pedrosa et al. (2022). Partial range of motion training elicits favorable improvements in muscular adaptations compared to full range of motion training. Eur J Sport Sci, 22(8), 1240–1250.
• Schoenfeld, B.J. (2010). The mechanisms of muscle hypertrophy and their application to resistance training. J Strength Cond Res, 24(10), 2857–2872.