REF-1166 · Grgic et al., 2022 · Biology of Sport · Paper · PubMed · Free full text

Is the mid-thigh pull a trustworthy strength test?

Across 16 studies peak force repeated well (median ICC 0.96, CV 4.9%). Mostly team-sport athletes; standardise the position. A change under ~5% is noise; only clearly larger ones count.

B+moderate evidenceMeasured on recreational to highly trained athletes

Grgic, Scapec, Mikulic and Pedisic searched five databases to December 2020 for studies reporting test-retest reliability of peak force in the isometric mid-thigh pull (IMTP), in which the athlete pulls maximally on a fixed bar at mid-thigh height. Sixteen studies (307 participants, 8 to 59 per study) qualified; 15 were rated excellent and one moderate on the COSMIN checklist.

ICCs ranged from 0.73 to 0.99 (median 0.96), with 78% at 0.90 or above; CVs ranged from 0.7% to 11.1% (median 4.9%), with 58% at 5% or below. Reliability held for bilateral and unilateral pulls, absolute and relative peak force, and across knee and hip angle variants as long as they were standardised.

Eight of ten studies that tested it found no systematic change between sessions. The authors conclude the IMTP is a reliable maximal-strength test for practice and research.

Included studies

Search: CINAHL, PubMed/MEDLINE, Scopus, SPORTDiscus and Web of Science, to 1 December 2020, plus reference lists; two reviewers independently.
Inclusion: peer-reviewed English studies reporting ICC and/or CV for absolute or relative peak force in the bilateral or unilateral IMTP.
Studies: 16 (15 from the primary search, 1 from reference lists); 307 participants, 8-59 per study (median 16); 15 reported ICCs, 14 CVs; 1-7 days between sessions.
Test: a maximal overcoming isometric pull on an immovable bar; knee angles mostly 120-150 degrees and hip angles 125-175 degrees (or self-selected); warm-up usually two attempts at 50% and 75% of perceived maximum; sampling 200-1000 Hz.
Quality: COSMIN form B (11 items); 15 studies scored 9 (excellent), one 8; no study met item 3 (sample size of at least 100).

Key Results

SubsetICC range (median)Share of ICCs ≥ 0.90CV range (median)Share of CVs ≤ 5%
All studies0.73-0.99 (0.96)78% (98% ≥ 0.75)0.7-11.1% (4.9%)58%
Bilateral IMTP (15 studies)0.73-0.99 (0.96)75%0.7-11.1% (5.3%)43%
Unilateral IMTP (2 studies)0.89-0.97 (0.94)88%3.4-4.9% (4.1%)100%
Absolute peak force (16 studies)0.84-0.99 (0.97)88%0.7-11.1% (4.6%)61%
Relative peak force (4 studies)0.73-0.94 (0.88)38%3.4-10.1% (5.7%)50%
Other findingResult
Systematic change between sessions (10 studies tested)8 none; 1 higher first session, 1 higher second (pre-peak-height-velocity girls); d 0.21-0.29
Nine knee-hip combinations (Comfort et al., 8 athletes)ICC 0.98-0.99 in every position
Least reliable absolute-force position120 degrees knee, 125 degrees hip, mid-thigh bar: ICC 0.84, CV 11.1% (Guppy et al.); the lowest ICC overall was 0.73, for relative force
Familiarisation10 studies had at least one session; 5 relied on prior experience; reliability similar

Mechanisms

Authors' interpretation: standardisation of joint angles matters for the force value (an upright position gives higher force than a bent one) but not much for reliability.
Inference: with a CV near 5%, a change under about 5% in one athlete is indistinguishable from test noise, and a change has to clearly exceed it to be trusted; a smallest detectable change of about 8.5% has been cited for IMTP peak force (REF-1167 introduction). The base's reading, not an analysis by these authors.

Practical Application

Is the mid-thigh pull a trustworthy strength test? Across 16 studies peak force repeated well (median ICC 0.96, CV 4.9%). Mostly team-sport athletes; standardise the position. A change under ~5% is noise; only clearly larger ones count.
Reliability held for bilateral and single-leg pulls; no study tested endurance athletes.
Use it to track whether a strength block worked without a 1RM. It is short, not very fatiguing and has a very low reported injury rate (authors' introduction). Whether IMTP gains predict endurance performance is a separate question (REF-1167).
Relative (per kg) force was less reliable (median ICC 0.88) than absolute force; track absolute force and body mass separately, the base's suggestion following this result.
One familiarisation session or a practice warm-up is enough for people who have done the test before; the authors still advise a separate familiarisation as optimal practice.

Limitations

Measurement studies were small (median n 16); none met the COSMIN sample-size item; only two stated which ICC type they used, and ICC types can differ widely on the same data.
Peak force only; reliability of rate of force development, impulse and time-specific forces is not covered.
Mostly team-sport and youth athletes; no endurance athletes, no older adults; only two studies of the unilateral test.
Descriptive summary without pooled estimates or heterogeneity analysis.
Funding: not stated in the full text read; no conflicts of interest declared.
How this entry was checked and changed
Why this level

systematic review of small test-retest measurement studies (median n 16, none met the COSMIN sample-size item), ICC types mostly unstated, no pooled estimate; reliability studies, not controlled trials, so below the A tier; read from the full text

Full text (checked 2026-10-16)
Sample sizes ranged from 8 to 59 participants (median = 16); overall number of participants 307; thirteen studies included athletes (full text).
ICCs ranged from 0.73 to 0.99 (median ICC = 0.96), 78% ≥ 0.90 and 98% ≥ 0.75; CVs 0.7% to 11.1% (median CV = 4.9%), 58% ≤ 5% (full text).
Bilateral: ICCs 0.73 to 0.99 (median 0.96), CVs 0.7% to 11.1% (median 5.3%), 43% of CVs ≤ 5% (full text).
Unilateral: ICCs 0.89 to 0.97 (median 0.94), CVs 3.4% to 4.9% (median 4.1%) (full text).
Absolute peak force: ICCs 0.84 to 0.99 (median 0.97); relative peak force: ICCs 0.73 to 0.94 (median 0.88), CVs 3.4% to 10.1% (median 5.7%) (full text).
Ten studies examined systematic changes; eight found none; differences small (Cohen's d: 0.21 to 0.29) (full text).
Comfort et al.: all ICCs for peak force ranged from 0.98 to 0.99 across nine knee and hip angle combinations (full text).
Fifteen studies classified as excellent methodological quality, one moderate; none satisfied item 3 (sample size) of the COSMIN checklist (full text).
No potential conflict of interest was reported by the authors (full text).