Posterior

What the evidence base looks like

The only glute-specific meta-analysis pooled 12 longitudinal studies with 318 participants, 144 of them women. Eight used untrained or detrained people and four used trained people. Six were rated good quality and six moderate. The pooled effect was moderate (SMD 0.71, 95% CI 0.50–0.91, from 11 studies, I² = 22%), with single exercises (0.74) and combined protocols (0.68) performing similarly.^[1] Trained and untrained subgroups showed similar effects, but only three trained studies could be pooled, and most ultrasound studies measured a single, upper glute site.

Why percentages can't be compared across studies. Studies differ in subjects, duration, volume, imaging method (thickness, cross-sectional area or whole-muscle volume) and the region measured. A 15% gain in one study does not beat a 9% gain in another. The charts in this report only compare conditions inside the same study.

Source labels used in this report

Peer-reviewed journal article · Preprint not yet peer reviewed · Abstract conference abstract with limited detail · First-party the practitioner's own content · Media reported by others · Unverified figures that couldn't be re-confirmed.

How this report was assembled

Four input reports were merged: a research-based movement selection pass, an independent literature review with evidence grades, and two independent practitioner surveys. Where they disagreed, both readings are shown. On September 29, 2026, the three newest trials were re-checked: the 2026 leg press study and the 2025 stretch-vs-squeeze preprint were confirmed in full; the 2026 extension-vs-abduction trial was confirmed as a Journal of Strength and Conditioning Research article, but its regional percentages couldn't be re-read from the full text and are marked approximate.

The three muscles

Glute max is the large, superficial muscle that drives hip extension and external rotation; its upper fibers also help abduct the hip. Glute med sits above and to the side, and glute min lies beneath it. Both abduct the hip and stabilize the pelvis on one leg.

Why regions respond differently

A 2025 torque study (cited via the meta-analysis) found the lower glute max produces more extension and external-rotation torque, while the upper region produces more abduction torque.^[22] Training only in the sagittal plane (squats, thrusts, hinges) therefore under-trains the upper glute.

The 2026 extension-vs-abduction trial tested this directly.^[4] After 10 weeks, cable hip extension grew the glute max about 9% overall with the largest change distally, while cable abduction concentrated growth proximally and produced a small but significant glute med/min increase. Strength gains were specific to the trained movement. The authors judged the two stimuli nonredundant. They used "shelf" and "teardrop" as communication terms but did not measure shape, so the finding is about regions, not a guaranteed look.

The studied techniques, as described in the independent literature review: extension with the trunk leaned forward about 70°, knee straight, moving from about 20° of hip flexion to full extension; abduction standing, knee straight, trunk neutral, starting from about 20° of cross-body adduction. One review also recorded roughly 10 sets per week at 8–12RM. Unverified These protocol details come from the study's published summary and couldn't be re-read from the full text.

Other extension work shows the same lower-biased pattern. In the hip thrust vs squat trial, both groups grew the lower glute max most, and glute med/min and hamstrings showed little growth in either.^[2,1] The 2026 authors noted that cable extension, hip thrusts and squats produce broadly similar distal-dominant growth. Abduction is what fills the upper gap.

Coaches have long made regional claims from EMG (Contreras: vertical work favors the lower glutes, abduction the upper glute and med; Nippard: machine abduction is his "best upper glute" pick).^[33,29,37] The abduction claim now has growth data behind it. Other regional claims remain unproven.

Cross-cutting findings

• Meta-analysis guidance. Prioritize single-joint work like the hip thrust when the goal is glute max growth without growing other muscles. Parallel or full-depth squats are effective, and the leg press and kneeling hip extension can also work. Squats and leg presses may not suit Wellness or Bikini competitors who must control quad size.^[1]
• Free weights vs machines. Across 13 studies and 1,016 participants (219 women), hypertrophy did not differ. Strength gains were specific to the modality trained, so numbers reset when equipment changes.^[10]
• Exercise variation. Some planned variation may enhance regional hypertrophy, while excessive random variation may compromise gains. All subjects were young men.^[11]

• The hip thrust paradox. The thrust produced higher glute EMG than the squat at every site measured, yet the two grew the glutes similarly.^[2]
• The depth paradox. Glute EMG drops below parallel and the glute's leverage is short there, yet deeper squats grew the glutes more.^[5,28]
• What force modeling suggests instead. In the same group of athletes, the top force tier for glute max was the loaded split squat, loaded single-leg RDL and loaded single-leg hip thrust. For glute med: side plank, loaded single-leg squat and single-leg RDL. For glute min: single-leg RDL and side plank. Banded side steps were not top tier for any glute muscle.^[9] "Trained" here meant at least three months of lower-body strength training. Force is not growth either, but it is a better proxy than EMG.
• EMG findings still cited by coaches. A wider squat stance raises glute max EMG.^[14] The seated abduction machine gave the best glute med to TFL activation ratio compared with side-lying abduction and clamshells (11 trained men).^[13] Step-ups, split squats and lunges reach very high glute activation (above 60% of maximum).^[12]

• Stretch vs squeeze (preprint, 2025). Twenty untrained people trained a Multi-Hip hip extension for 10 weeks, with each limb randomized to a cam that peaked when the glute was long or short. Glute max volume grew +5.8% vs +6.5% (P = .51), with no difference in any muscle.^[15] It is not yet peer reviewed, and Ethier and Contreras funded it.
• Lengthened partials. In trained lifters, partial reps in the stretched half of the range matched full range.^[7] A 2023 meta-analysis hinted at a small benefit (SMD −0.28), but the confidence interval crossed zero.^[17] One conference abstract on hip extension favored long-length partials (+6.8% vs +3.1% hip extensors).^[16]
• What coaches say. RP: "stretch a ton," maximizing the angle between pelvis and thigh. Contreras and Ethier: pair one stretch-focused exercise with one squeeze-focused exercise. Nippard: train glutes both at long lengths (squats) and mid-range (thrusts). Carroll: pick exercises that are hard in either the lengthened or shortened position.^{[35,32,37,42]}
• Reconciled. Pairing a stretch-biased and a squeeze-biased movement is a sensible hedge and adds variety, but the only direct test found no advantage for either emphasis. Lengthened partials are optional, for example as extra reps after full-range failure.

The spectrum chart is an educated guess from lever mechanics. Cable exercises in particular shift with pulley height and body position.

• No glute dose-response research. The NSCA author noted none existed at the time of writing, and none turned up through September 2026.^[41]
• RP: start a block near the minimum effective volume and add sets only when performance, pump, soreness and recovery say more is productive. The 30–40+ prioritized ceiling is exceptional, not a starting target. Quad and hamstring training counts toward both the glute stimulus and the fatigue budget.^[35] Older RP material put the direct-work minimum at zero; a third-party summary lists 2017 figures of MEV 0 and MRV 16+.^[36]
• Contreras: his 36-set model splits roughly into 24 glute max-oriented and 12 glute med/min-oriented sets, with plenty of overlap. If training only brutal compounds to failure (squats, deadlifts, walking lunges), about 10 sets a week is the ceiling.^[30,34] He also says women generally recover better than men and can handle more volume (practitioner claim).^[31]

• Contreras: two sessions is a safe default and three suits specialization. In 2023 he called three the "sweet spot," with a practical range of two to six depending on genetics, exercise selection, load and effort, and he has said top Bikini pros train glutes three times a week.^[30,33,34]
• RP: two to five sessions, with three to four needed to reach specialization volumes because of the per-session ceiling.^[35] Henselmans: women prioritizing the lower body should train it at least twice a week.^[40]
• Recovery shapes frequency: lunges and other lengthened compounds cause the most damage, thrusts are intermediate, and abduction recovers fastest and can be trained most often.^[30]

RP's useful range runs 5–30 reps (about 30–85% of 1RM), with a starting mix of roughly half the sets at 10–20 and the rest split between 5–10 and 20–30. Deadlifts belong at 5–10 only, heavy spinal-loaded compounds should avoid 20–30, lunges work well at 20–30 but poorly at 5–10, and hip thrusts work in every range, with 10–20 best for most.^[35]

• RP: hold reps steady and add load. If you can't match previous reps two sessions in a row, the muscle has hit its local recovery limit: do a half session, then resume at the midpoint set count at about 2 RIR. Deload by dropping sets to the minimum, halving reps, and keeping week-1 loads for the first half of the week before halving them.^[35]
• Contreras: about 2 RIR on repeatable work; the first movement of a session can be pushed hardest.^[30] Ethier takes at least one hip thrust set to failure; Matthews keeps most sets close to failure; NSCA leaves a few reps in reserve on most sets and uses failure strategically.^[39,44,41]

Open or unverified items

• The extension vs abduction trial's regional percentages and protocol details couldn't be re-read from the full text.
• The two literature reviews describe the Short 2021 blood-flow-restriction result differently.
• RP's landmark figures differ between its 2024 guide and older material.
• Stern's month-3 frequency is ambiguous (three sessions stated, four day types listed).
• The NSCA article's author is inferred; the Chapados source is sponsor content; the Perform podcast date is unconfirmed; the Cela and some Contreras details rely on third-party summaries.
• Nippard's walking lunge is described as S+ in one summary and S in another.

Not covered by these passes

Sex-specific hypertrophy responses, the general (non-glute) volume dose-response literature, research on exercise order, and nutrition.