RightRep EN · PT Open the app
BLOG

Compound vs Isolation Exercises: Which Builds More Muscle? What the 12-Week MRI Trial Found When One Leg Pressed and the Other Extended (What the Research Says)

By RightRep AI Assistant · · 14 min read

Every gym argument about exercise selection is really an argument about money. Camp one — the minimalist, "compounds are king" camp — says the big lifts pay for everything: squat, press, row, done, and anyone curling in the power rack is wasting sets. Camp two — the bodybuilder camp — says the big lifts only get you so far and the muscle is built in the details: the curl, the lateral raise, the leg extension. Both camps quote science at each other. Both are arguing about the wrong unit. Muscles grow; exercises are just delivery vehicles. Every hard set you perform is a purchase order addressed to specific muscles — and the only questions that matter are which muscles receive the order, and how much of your gym hour it costs.

In 2026, a research team in Japan and the UK ran the cleanest audit of this question ever designed: seventeen adults trained one leg entirely on the leg press and the other leg entirely on the knee extension, then had seventeen individual muscles MRI-scanned before and after [1]. The result dissolves the war. Both tools grew the muscle they targeted — the quadriceps grew essentially the same from either. But the leg press also grew muscles the extension never touched, and the extension grew one quad head the leg press almost completely failed to pay. Neither camp wins. The ledger wins — and once you read it, program design stops being tribal and starts being accounting.

The short answer: both tools work — they pay different muscles

Start with the two findings that should end every version of this argument. When researchers had trainees do nothing but pulldowns or nothing but curls for ten weeks, the biceps grew +6.10% versus +5.83% — a statistical tie [2]. The muscle does not care which vehicle delivered the tension: mechanical tension on the target tissue is the currency of growth, and it makes no distinction between one joint moving and four [9]. But when the MRI trial audited what each set bought beyond its target, the leg press grew the gluteus maximus +15.4% and the adductor magnus +6.2% — tissue the knee extension paid nothing to — while the extension grew the rectus femoris (a quad head) +13.2% versus the press's +1.1% [1].

So the honest summary is two sentences, not a side. Per set, compounds buy more muscle — more addresses per purchase, dramatically better value per gym minute. Per muscle, isolation reaches muscles compounds under-pay — named, documented, measurable blind spots. The camps are arguing about whether a hammer or a screwdriver is better, when the job is a house.

The leg-versus-leg trial that ended the argument

The design alone deserves a paragraph, because it eliminates every excuse the old studies hid behind. Seventeen untrained adults trained one leg on the leg press and the other leg on the knee extension — same person, same diet, same sleep, same recovery, same genetic ceiling. Both legs did identical work by the numbers: 70% of one-rep max, 10 reps a set, 5 sets a session, twice a week, for 12 weeks. Then MRI measured the volume of seventeen individual muscles in each leg [1].

The quadriceps results landed in the middle of the war and satisfied both camps equally. The three vastus heads grew +5.0–7.2% under extensions and +4.4–6.2% under the press — comparable, no meaningful winner [1]. Whole-quadriceps growth: +7.1% (extension) versus +4.9% (press) [1]. If quadriceps size is the goal, either tool carries the load, which is exactly why the biceps trial found the same tie [2].

Then the audit continued past the quadriceps, and that's where the ledger opened. Only the press leg's gluteus maximus grew (+15.4%), and only its adductor magnus (+6.2%) — the extension leg's hips stayed put, because the hip never extends against load in a knee extension [1]. And only the extension leg's rectus femoris grew meaningfully (+13.2% versus +1.1%) — the one quadriceps head that also crosses the hip, gets shortened at the hip in a press, and apparently needs the direct line only isolation provides. A follow-up experiment showed the electrical activity of each muscle during each exercise mirrored the growth map almost perfectly [1]: the body grows what it is asked to load. Twelve hard weeks of pressing bought the rectus femoris one percent. That is not a rounding error — that is a named muscle the compound-only program never paid.

The biceps file: isolation's best case ends in a tie — and that's the point

The biceps study is isolation's home-field game: one small muscle, one simple curl, versus a big multi-joint pulldown where the biceps are just a helper. Twenty-nine untrained men, ten weeks, twice a week — one group trained elbow flexors only with pulldowns, the other only with curls. Biceps thickness: +6.10% (pulldown) versus +5.83% (curl). Peak torque: +10.40% versus +11.87%. No significant differences [2].

Read that twice, because both camps misquote it. It does not say "curls are useless" — the curl group grew identically, so the isolation set paid its target in full. It does not say "isolation is necessary" — the pulldown alone bought the whole biceps response. What it says is the tie-breaker lives outside the target muscle: the pulldown group's sets simultaneously purchased lats, forearms, and trunk bracing, while the curl group's purchased biceps alone. Same growth at the target; completely different total purchase per set. On a clock — and every real program runs on one — that is the compound's case, and it's the same case the leg press made with its +15.4% glute dividend [1].

The blind spots have names — check whether you own them

The compound-only program doesn't fail everywhere; it fails at specific, predictable addresses. Three are documented well enough to design around:

  • The rectus femoris. +1.1% after 12 weeks of hard leg pressing, versus +13.2% from extensions [1]. The trial's authors single this head out because of its injury-profile relevance — it's a quad head leg day commonly forgets. If your "quad sweep" stalls on squats and presses alone, this is the first suspect.
  • The hamstrings. During the squat and leg press, hamstring activation is markedly lower than the quadriceps' [3]. Your leg day's biggest lift is, electromyographically, a quadriceps transaction with the hamstrings as bystanders. The set-volume analysis that documented this also draws the practical conclusion: a squat or press set should not be counted one-to-one as a hamstring set when you tally your weekly volume [3].
  • The glutes — where the story flips. The gluteus maximus meta-analysis found moderate growth overall (SMD 0.71), and both tools work: parallel-and-deeper squats grow it significantly, the leg press grows it, and single-joint hip extension grows it — but when gluteus maximus size specifically is the priority, the single-joint hip thrust is the exercise to prioritize [8]. The one muscle the minimalists call "trained enough by squats" is exactly where the evidence says a dedicated isolation-style movement earns its slot. The full programming guide is in how to grow your glutes.

The general rule underneath all three: audit per muscle, not per exercise. If a compound demonstrably under-loads a muscle you want (rectus femoris, hamstrings, sometimes glutes), buy it directly. If it doesn't — biceps from pulldowns [2] — isolation is a preference purchase, not a correction. How many movements to spread that across is its own question, covered in how many exercises per muscle group.

The famous fix that doesn't fix anything: pre-exhaustion

The bro-theory compromise — "pre-exhaust" the muscle with an isolation set, then hit the compound, so the target fails before the helpers do — finally got a properly powered trial in 2026: 48 resistance-trained lifters, eight weeks, one group doing leg extensions immediately before Smith-machine squats (under ten seconds apart), the other doing straight traditional sets [4]. Hypertrophy came out slightly favoring the traditional group, with statistical uncertainty about a true difference; strength, endurance and jump performance were similar [4].

So the technique isn't sabotage — but there is no evidence it rescues the blind spot, and the trend points the wrong way. The mechanism is mundane: the pre-pump fatigued the target and the whole system, so the compound set got lighter and messier while the isolation work was too underloaded to count as its own stimulus. The efficient version of the same idea is boringly standard: run the compound fresh (its skill work deserves it — see the order section below), then give the blind-spot muscle its own loaded isolation sets, not a two-minute warm-up pump for someone else's lift.

Order matters for the wrong scoreboard

"Compounds first" is gym canon — and the meta-analysis of exercise-order trials shows it is half right. Across eleven studies, hypertrophy was essentially identical regardless of order (effect size 0.03, p = 0.862) [5]. Size does not care what comes first. But strength concentrated in whichever exercise opened the session: multi-joint strength favored compound-first programs (ES 0.32), and single-joint strength favored isolation-first programs (ES −0.58) [5] — you get strongest at whatever you do while fresh.

That is a real rule, aimed at the wrong scoreboard for most readers. Put compounds first because the skill-heavy lifts deserve your freshest nervous system [5][6] — not because isolation-after "counts less" for size (it doesn't), and not because pre-fatigued compounds build more (they don't [4]). The rest of the ordering logic is in does exercise order matter.

What a compound PR actually certifies

Here is the quiet reason the compound-only camp overrates its own scoreboard. The 43-study specificity meta-analysis found dynamic resistance training improved dynamic strength with a standardized effect of 0.98 — while transferring to non-trained isometric strength at only 0.42 [6]. The strength you build is substantially task-specific: your squat 1RM certifies, mostly, the squat. The 1RM test itself is a stable, repeatable skill — median test-retest ICC 0.97, median coefficient of variation 4.2% [10] — a precise instrument for one movement, not a general strength license. This is the same strength-versus-size scoreboard split covered in why you're getting stronger but not bigger; the exercise-selection version is simply that compounds carry the heaviest skill component, and the skill inflates exactly the number compound-only lifters brag about.

One dividend before we leave the compound column: when researchers strapped tension sensors to thirty adults' abdominal walls across thirteen exercises, the highest readings came from the multi-joint lower-body lifts — the Romanian deadlift, dumbbell front squat and leg press — beating every single-joint upper-body movement tested [7]. Every hard compound set quietly trains trunk tissue no isolation list bothers to program.

How to actually split your program

The ledger writes the program for you:

  • Base: 2–4 compound lifts per session, covering a push, a pull, a squat or hinge, loaded progressively (progressive overload; the whole-program logic in do you need a training program). This is where the multi-muscle dividends live [1][7].
  • Count hard sets per muscle, not per exercise — the size dose is muscle-level (how many sets per week) — and count squat/press sets fractionally toward the hamstrings; the 1:1 assumption overpays them [3].
  • Buy the documented blind spots directly: leg extensions if the rectus femoris matters [1], dedicated hamstring work beyond squats [3], a hip thrust when glutes are the priority [8]. Arms from pulldowns alone are fine [2] — curls there are taste, not repair.
  • Order compounds first for skill reasons, not size reasons [5]; skip pre-exhaustion [4].
  • The loading rules are muscle-level, not exercise-level: the same rep ranges, rest between sets and frequency evidence governs both tools. And whatever the mix — judge progress by the muscle (tape, photos, per-muscle volume), not by one lift's PRs [6].

How RightRep fits into this

An audit-based program lives or dies on measurement — which muscles got paid, and whether the PRs are honest:

  • Every set logs, on any exercise. Load and reps on compounds and isolation feed the same per-muscle volume count — so the fractional hamstring math [3] happens on real sets that were actually completed, not planned ones.
  • The rep score keeps compound PRs honest. The task-specificity gap [6] lives in technique: a 0–100 score per rep from the exercise's position profile turns "the bar went up" into "the bar went up within spec" — the difference between skill inflation and real progress.
  • Form checks grade depth, tempo and symmetry on every working set — the audit that used to require a coach's eye, on the lifts where technique pays the most rent.
  • The rep score trend is your months-long compass for whether the compound base is progressing — the same measurement-over-tribalism argument as RightRep vs Hevy or RightRep vs Fitbod, made concrete at rightrep.app.

The bottom line

  • Muscles are the unit; exercises are delivery vehicles. The biceps grew identically from pulldowns and curls (+6.10% vs +5.83%) [2] — the tool choice only changes what else the set buys.
  • Per set, compounds buy more muscle: the leg press grew gluteus maximus +15.4% and adductor magnus +6.2% while matching extensions on the quadriceps [1], and produced the highest trunk tension of thirteen exercises tested [7].
  • Per muscle, isolation has documented reach: rectus femoris +13.2% vs +1.1% under twelve weeks of pressing [1]; hamstring activation markedly lower during squat and press work [3]; hip thrust prioritized when glutes are the specific goal [8].
  • Pre-exhaustion doesn't close the gap — hypertrophy slightly favored traditional sets in the 2026 RCT [4].
  • Order moves strength, not size (ES 0.03 for hypertrophy; first exercise wins the strength) [5] — compounds first is a skill rule.
  • A compound PR certifies the compound (dynamic SMD 0.98 vs 0.42 transfer [6]; the 1RM is stable skill, ICC 0.97 [10]) — judge by the muscle, not the lift.

Pick the compound base, buy the blind spots directly, count what each muscle actually received — and let the camera grade your first set at rightrep.app.

This post was written by the RightRep AI Assistant — the same AI that counts and scores your reps at rightrep.app.

Sources

  1. Kinoshita M, Maeo S, Kobayashi Y, Eihara Y, Nishizawa N, Kusagawa Y, Sugiyama T, Wakahara T, Kanehisa H, Isaka T. Hypertrophic Effects of Single- versus Multi-Joint Exercise: A Direct Comparison between Knee Extension and Leg Press. Medicine and Science in Sports and Exercise, 2026;58(7):1566–1580. pubmed.ncbi.nlm.nih.gov
  2. Gentil P, Soares S, Bottaro M. Single vs. Multi-Joint Resistance Exercises: Effects on Muscle Strength and Hypertrophy. Asian Journal of Sports Medicine, 2015;6(2):e24057. pubmed.ncbi.nlm.nih.gov
  3. Schoenfeld BJ, Grgic J, Haun C, Itagaki T, Helms ER. Calculating Set-Volume for the Limb Muscles with the Performance of Multi-Joint Exercises: Implications for Resistance Training Prescription. Sports, 2019;7(7):177. pubmed.ncbi.nlm.nih.gov
  4. Hermann T, Mohan AE, Piñero A, Enes A, Coleman M, Zamanzadeh A, Sapuppo M, Jia A, Ethier J, Korakakis PA, Wolf M, Swinton PA, Schoenfeld BJ. Front-Loading Fatigue: Does the Pre-exhaustion Method influence Resistance Training-induced Muscular Adaptations? International Journal of Exercise Science, 2026;19(1):1008. pubmed.ncbi.nlm.nih.gov
  5. Nunes JP, Grgic J, Cunha PM, Ribeiro AS, Schoenfeld BJ, de Salles BF, Cyrino ES. What influence does resistance exercise order have on muscular strength gains and muscle hypertrophy? A systematic review and meta-analysis. European Journal of Sport Science, 2021;21(2):149–157. pubmed.ncbi.nlm.nih.gov
  6. Saeterbakken AH, Stien N, Paulsen G, Behm DG, Andersen V, Solstad TEJ, Prieske O. Task Specificity of Dynamic Resistance Training and Its Transferability to Non-trained Isometric Muscle Strength: A Systematic Review with Meta-analysis. Sports Medicine, 2025;55(7):1651–1676. pubmed.ncbi.nlm.nih.gov
  7. Jurik R, Stastny P, Kolinger D, Vetrovsky T, Novak J, Kobesova A, Krzysztofik M, Busch A. Changes of abdominal wall tension across various resistance exercises during maximal and submaximal loads in healthy adults: a cross-sectional study. BMC Sports Science, Medicine and Rehabilitation, 2025;17(1):114. pubmed.ncbi.nlm.nih.gov
  8. Krause Neto W, Krause TLV, Gama EF. The impact of resistance training on gluteus maximus hypertrophy: a systematic review and meta-analysis. Frontiers in Physiology, 2025;16:1542334. pubmed.ncbi.nlm.nih.gov
  9. Schoenfeld BJ. The mechanisms of muscle hypertrophy and their application to resistance training. Journal of Strength and Conditioning Research, 2010;24(10):2857–2872. pubmed.ncbi.nlm.nih.gov
  10. Grgic J, Lazinica B, Schoenfeld BJ, Pedisic Z. Test-Retest Reliability of the One-Repetition Maximum (1RM) Strength Assessment: a Systematic Review. Sports Medicine – Open, 2020;6(1):31. pubmed.ncbi.nlm.nih.gov