Healthy lower-limb tendons stiffen with loading (stiffness SMD 0.74, modulus 0.82, cross-section only 0.22 across 61 studies), the gain is mostly a material change, resistance training works best, and protocols that reached high tendon strain (~5%) beat low-strain ones (~3%)
A contrast that rests on three high-strain intervention groups from the few studies that measured strain
Lazarczuk and colleagues pooled 61 training studies (763 people) that measured lower-limb tendon stiffness, Young's modulus or cross-sectional area (CSA) before and after a loading programme, mostly in the Achilles (33 studies) and patellar tendon (24). Loading raised stiffness moderately (SMD 0.74), modulus largely (0.82) and CSA only a little (0.22).
In the 20 studies that measured all three, the stiffness gain tracked the modulus gain (p < 0.001) and not the CSA gain (p = 0.19): tendons get stiffer mainly by changing their material, not by growing.
Resistance training was the only training type that raised all three properties (stiffness SMD 0.80); jump training raised stiffness less (0.49), and the two concurrent resistance-plus-aerobic groups showed no change (0.03).
Within resistance training, relative intensity (≥70% vs <70% of 1RM or MVC), volume, duration (≥12 vs <12 weeks) and contraction mode did not significantly change the stiffness response.
The one loading parameter that did was tendon strain: protocols reaching about 5% strain produced large gains in stiffness (SMD 1.04) and modulus (0.82), while protocols at about 3% produced no significant change. Only three included studies reported strain; the Discussion credits the finding to two strain-controlled calf studies from one laboratory (REF-1153 is one of them).
Adults responded more than elderly people.
Included studies
Search: PubMed, Scopus, CINAHL, SPORTDiscus and EMBASE; healthy people; loading interventions of at least 2 weeks with loading parameters reported; in vivo tendon stiffness, modulus or CSA.
Studies: 61 articles, n=763 (615 men, 135 women, 13 not specified); 46 resistance-only, 9 jump-only, 3 resistance vs jump, 1 aerobic (cycling), 2 concurrent resistance plus aerobic. Achilles 33 studies, patellar 24.
Resistance-training groups by contraction mode: 22 concentric-eccentric, 19 isometric-only, 9 eccentric-only, 4 concentric-only. Isometric protocol intensity ranged from 50 to 90% of 1RM/MVC; hold durations varied and were not matched (the authors list this as a limitation).
Isometric studies relevant here (from the review's table): Arampatzis 2007 (14 weeks, 4×/week, low- vs high-strain isometric plantarflexion, REF-1153); Kubo 2001 (12 weeks, 4×/week, short vs long isometric knee-extension contractions, REF-1154); Albracht 2013 and Bohm 2021 (14 weeks, 4 and 3-4×/week, high-intensity low-volume isometric plantarflexion, REF-1155, REF-1156); Fletcher 2010 (8 weeks, 3×/week, isometric plantarflexion in 6 highly trained runners).
Definitions: high intensity ≥70% of 1RM or MVC; high strain ~5%, low strain ~3%; low volume ≤3100 arbitrary units.
Analysis: Hedges' g pre-to-post SMD per group, DerSimonian-Laird random effects; subgroup analyses of age, duration and contraction mode restricted to high-intensity resistance groups; matched-study analyses where available; meta-regression of stiffness change on modulus and CSA change. Quality by a revised PEDro scale; small-study bias by Egger's test.
Key Results
| Comparison | Stiffness SMD (95% CI) | Modulus SMD (95% CI) | CSA SMD (95% CI) |
|---|---|---|---|
| All loading | 0.74 (0.62-0.86), 81 groups | 0.82 (0.58-1.07), 38 groups | 0.22 (0.12-0.33), 68 groups |
| Resistance training | 0.80 (0.67-0.94), 68 groups | 0.90 (0.65-1.15), 35 groups | 0.23 (0.12-0.34), 59 groups |
| Jump training | 0.49 (0.22-0.77), 11 groups | no clear change | no clear change |
| Concurrent resistance + aerobic | 0.03 (-1.03 to 1.10), 2 groups | no change | no change |
| High intensity (≥70%) | 0.86 (0.70-1.25), 53 groups | 0.91 (0.63-1.18), 25 groups | 0.22 (0.09-0.35), 43 groups |
| Low intensity (<70%) | 0.74 (0.16-1.31), 5 groups | 0.95 (0.23-1.67), 8 groups | 0.10 (-0.21 to 0.41), 8 groups |
| High strain (~5%) | 1.04 (0.65-1.43), 3 groups | 0.82 (0.44-1.20), 3 groups | 0.25 (-0.11 to 0.61), 3 groups |
| Low strain (~3%) | not significant | not significant | not significant |
| High vs low strain | p = 0.006 (χ2, n = 7 groups); p = 0.007 in the abstract | p = 0.009 | — |
| ≥12 vs <12 weeks (high intensity) | 0.88 vs 0.78, no difference | 0.80 vs 1.48, no difference | 0.16 vs 0.49 |
| Adults vs elderly (high intensity) | 0.91 vs 0.40, p = 0.02 | 1.05 vs 0.21, p = 0.002 | 0.24 vs 0.06 |
Meta-regression (20 studies): modulus change was the main moderator of stiffness change (p < 0.001); CSA change was not (p = 0.19).
Contraction mode: all modes raised stiffness (SMD 0.62-1.14) and modulus (0.74-1.27) with no significant difference between modes; only eccentric training raised CSA (0.58, 95% CI 0.22-0.93, 6 groups), and that disappeared in the matched analysis.
Volume: low 0.81 vs high 0.85 for stiffness, no difference.
Quality: PEDro median 4 of 8 (IQR 3-6); allocation concealment reported in 7% of studies, assessor blinding in 18%.
Mechanisms
Measured: stiffness gains are explained by modulus (material) gains, not by tendon growth (meta-regression).
Measured, thin: high-strain protocols gave larger stiffness and modulus gains than low-strain protocols (three high-strain groups).
Authors' interpretation: there may be a strain threshold for adaptation, consistent with cell studies in which tenocytes respond to larger cyclic strains. Longer contractions at a given load may transmit more strain to tendon cells because tendon is viscoelastic, which could explain why sustained loading beat jump training.
Authors' interpretation: aerobic training done alongside resistance training may blunt tendon adaptation, as concurrent training blunts muscle gains (two groups only; mechanism unknown).
Inference: relative intensity did not separate protocols because %MVC or %1RM is a poor proxy for the strain the tendon actually sees; the authors make the same point.
Practical Application
Want stiffer tendons? In 61 studies (763 mostly untrained people), resistance training raised tendon stiffness (SMD 0.80); protocols straining the tendon ~5% did best (SMD 1.04), but that rests on just 3 training groups. Contraction mode, volume and 12+ vs under 12 weeks made no clear difference.
The authors' prescription when strain cannot be measured: use a very high relative intensity (about 90% or more of 1RM or MVC), as in the high-strain protocols. Isometric holds are one way to reach it; the review found no contraction mode better than another.
Endurance athletes, expect less. The two concurrent strength-plus-aerobic groups showed no stiffness change, and the three studies of trained athletes (highly trained distance runners, female endurance athletes, strength-trained men) showed no tendon stiffness change after high-intensity training. TRANSFER from untrained people; the gain may be smaller in a runner or cyclist.
Timing: in the three studies with repeated measurements, stiffness changes showed between 4 and 8 weeks. Not from this paper; the base's practice: judge a tendon-loading block after at least 8 weeks and keep it going through the season rather than stopping and starting.
Limitations
Effects are within-group pre-to-post SMDs, not differences against a control group, because not all studies had a control and controls differed (active vs passive). Familiarisation and measurement drift are not removed.
Mostly small studies (all but one with 15 or fewer participants), median PEDro 4 of 8; allocation concealment and assessor blinding rarely reported; randomisation not reported in 21% of studies, and single-group designs were included.
Strain was reported in only three included studies; the Discussion's "strong evidence" for high strain rests on three high-strain intervention groups, and the two studies it names are calf studies from one laboratory (REF-1153 and its 2010 sequel).
Isometric protocols were grouped regardless of hold duration and time under tension.
Tendons and aponeuroses, Achilles and patellar, and different measurement methods were pooled.
Interventions lasted at most 14 weeks except one; long-term adaptation is unknown.
Some small-study bias on Egger's test; the authors report that adjusted SMDs did not change the conclusions.
Funding: none for the review ("No sources of funding were used to assist in the preparation of this article"); open access funded by CAUL. The authors declare no conflicts of interest.
How this entry was checked and changed
Why this level
meta-analysis of mostly small studies (all but one had 15 or fewer participants), effects are pre-to-post changes rather than differences against a control group, median PEDro score 4 of 8, and the headline high- vs low-strain contrast rests on 3 vs 4 intervention groups; read from the full text
Full text (checked 2026-10-16)
Sixty-one articles, 763 participants; Achilles tendon 33 studies, patella tendon 24 studies; resistance training 49 studies (full text).
Stiffness SMD 0.74 (95% CI 0.62-0.86; 81 intervention groups), modulus SMD 0.82 (95% CI 0.58-1.07; 38 intervention groups), CSA SMD 0.22 (95% CI 0.12-0.33; 68 intervention groups) (full text).
Meta-regression of 20 studies: modulus the predominant moderator of stiffness (p < 0.001); CSA not significant (p = 0.19) (full text).
Resistance training stiffness SMD 0.80 (95% CI 0.67-0.94; 68 intervention groups); jump-based SMD 0.49 (95% CI 0.22-0.77; 11 intervention groups); concurrent SMD 0.03 (95% CI -1.03-1.10; two intervention groups) (full text).
High-strain (~ 5%) stiffness SMD 1.04 (95% CI 0.65-1.43; three intervention groups) and modulus SMD 0.82 (95% CI 0.44-1.20); χ2 (1, n = 7) = 7.45, p = 0.006 for stiffness and 6.76, p = 0.009 for modulus (full text).
High-intensity stiffness SMD 0.86 (95% CI 0.70-1.25; 53 intervention groups); low-intensity SMD 0.74 (95% CI 0.16-1.31; five intervention groups) (full text).
19 groups underwent isometric-only protocols; isometric protocol intensity ranged from 50 to 90% 1RM/MVC (full text).
Adults stiffness SMD 0.91 vs elderly 0.40 (p = 0.02); modulus 1.05 vs 0.21 (p = 0.002) (full text).
PEDro median score of 4 (IQR: 3-6); allocation concealment reported in 7% of studies, assessor blinding in 18% (full text).
No sources of funding were used to assist in the preparation of this article; no conflicts of interest (full text).