Loading a tendon and remodeling a tendon are two different things, and the imaging literature is unusually clear about the distance between them. Loading is the mechanical stimulus: how much the tissue is stretched, at what rate, for how long. Remodeling is the structural response: does the tissue get bigger, does its material change, does it behave differently under the same test. Most arguments about tendon work collapse because the two get used as synonyms. Here is what the imaging actually records when you take a healthy tendon and look at it properly.
Start with the number that should end the argument
If you want to know how much the Achilles tendon itself stretches, modern three-dimensional MRI is the cleanest answer available, and the answer is small. A 2026 study scanned eighteen healthy adults (ten women, eight men, mean age twenty-four) at three ankle angles — twenty degrees of plantar flexion, neutral, and twenty degrees of dorsiflexion — and computed tendon length in three dimensions rather than two (J Anat, PMID 41403356). The free Achilles tendon elongated by 1.2% ± 0.3 across that whole range of movement.
That is the entire free tendon. Anything a person feels stretching in the back of the calf is mostly the muscle, the muscle-tendon junction, and the aponeurosis, not the tendon substance itself. This matters for load prescription, because if you are chasing a strain figure you can hit, one per cent is not much room, and a training plan built around a number you cannot reach is a plan built on a misreading.
The same study is a warning about the measurement
The more useful half of that 2026 paper is what happened when the authors did it the ordinary way. The conventional two-dimensional method failed to detect physiologically relevant strain, and in places indicated an apparent shortening of the tendon — the tissue appearing to get shorter as the ankle moved towards a position that should lengthen it. The authors then ran a simulation: introduce a sagittal-plane misalignment of two to four millimetres, which is a very ordinary error when someone is marked up in a scanner, and the measured tendon length moved by 2.8 to 6.3 millimetres.
Read that twice. A few millimetres of marking error, in a tendon whose real length change is around one per cent, produces swings larger than the effect people are trying to detect. This is not a fringe observation. It is the reason a scan in one clinic and a scan in another can disagree about a tendon that never changed. Two-dimensional ultrasound and MRI are excellent at seeing a tendon and poor at being trusted with a length.
Loading reliably changes behaviour; remodeling is the flaky half
The experimental evidence is cleaner when researchers control one leg against the other, because it removes the guesswork about how hard someone trained.
Eleven men trained for fourteen weeks, four days a week, with one leg at a low tendon strain of 2.97% ± 0.47 and the other at a high strain of 4.72% ± 1.08, at the same loading frequency and the same exercise volume (J Biomech, PMID 20863501). Ultrasound, MRI and a dynamometer before and after found that only the high-strain leg showed a decrease in strain at a given tendon force, and increases in tendon-aponeurosis stiffness and elastic modulus. The cross-sectional area showed no statistically significant difference in either leg. The tendon got stiffer where it was asked to be stiffer; it did not get measurably bigger anywhere.
A second study from the same group, fourteen weeks again, pushed the strain to about 6.5% and varied the shape of the loading rather than just the amount (J Exp Biol, PMID 25267851). Here the remodeling half does show up: cross-sectional area rose by 4.2% in the reference protocol and 5.3% in the long-duration protocol, tendon stiffness rose by 57% and 25%, and Young’s modulus by 51% and 17%. High-rate training produced region-specific hypertrophy, while the increase in the reference leg was only a tendency.
Read those two studies next to each other and a usable pattern appears. Behavioural change — stiffness, modulus, strain at a given force — shows up in the high-strain leg at around three to five per cent of tendon strain, reliably enough to be worth designing around. Structural change, measured as cross-sectional area, is smaller, more variable, and absent in at least one of the two experiments. The strain magnitude is the lever. The rate and the duration are refinements, not the main event.
Why the published numbers are hard to compare
A 2023 survey screened 107 articles reporting in-vivo Achilles strain in healthy adults during walking, running, jumping, contractions and landings (Sports Med Open, PMID 37466866). Its finding is not a new strain value; it is a warning. Because anatomical definitions of the sub-tendon and the aponeurosis are inconsistent across studies — where a study stops measuring, and whether the muscle-tendon junction is inside or outside the segment — the literature contains considerable confusion, and a value from one paper cannot simply be stacked on a value from another. MRI, ultrasound and motion capture were the dominant approaches, often with modelling on top.
That survey is the reason this article quotes strain percentages with an asterisk of caution. The intervention studies above are stronger evidence precisely because they measured the same tissue in the same person on both sides with the same protocol, rather than comparing numbers across papers.
Where experimental biology fits, and where it does not
A recurring claim is that healing-promoting research peptides are the shortcut past the boring mechanics above. The honest position is that the clinical evidence for those compounds in tendon injury is a separate question, and a thinner one; the load-capacity models and the imaging literature say nothing about them either way. For anyone following the published work rather than the marketing, the tendon imaging literature is the reference point: the material properties, the strain-by-load relationship, and the measurement error floors.
What this does not settle
- One tendon. Everything quoted here is the Achilles in healthy adults aged roughly twenty to fifty. Patellar and rotator cuff tendons load differently and do not have these numbers.
- Fourteen weeks. No trial here ran for years, and tendon adaptation continues well past the study window.
- Healthy tissue. A tendon that has been injured does not remodel the way an intact one does, and none of these cohorts included injured participants.
- Imaging endpoints, not outcomes. Cross-sectional area and modulus are structural measurements. Whether any of it changes how someone jumps, trains or avoids injury is a separate question that these studies did not answer.
- No blinding to the fundamental confound. Paired-leg designs control the stimulus beautifully and cannot control for who trained harder on their assigned leg.
The catalogue row for the compound most often brought up in tendon discussions sits here: the BPC-157 product page. It is a catalogue entry, not a finding. None of the imaging studies above tested BPC-157 in people, and none of them measured tendon remodelling under load with it.
