1. The objectives of this study were to (1) quantify experimentally in vivo changes in pennation angle, fibre length and muscle thickness in the triceps surae complex in man in response to changes in ankle position and isometric plantarflexion moment and (2) compare changes in the above muscle architectural characteristics occurring in the transition from rest to a given isometric plantarflexion intensity with the estimations of a planimetric muscle model assuming constant thickness and straight muscle fibres. 2. The gastrocnemius medialis (GM), gastrocnemius lateralis (GL) and soleus (SOL) muscles of six males were scanned with ultrasonography at different sites along and across the muscle belly at rest and during maximum voluntary contraction (MVC) trials at ankle angles of -15 deg (dorsiflexed direction), 0 deg (neutral position), +15 deg (plantarflexed direction) and +30 deg. Additional images were taken at 80, 60, 40 and 20% of MVC at an ankle angle of 0 deg. 3. In all three muscles and all scanned sites, as ankle angle increased from -15 to +30 deg, pennation increased (by 6-12 deg, 39-67%, P < 0.01 at rest and 9-16 deg, 29-43%, P < 0.01 during MVC) and fibre length decreased (by 15-28 mm, 32-34%, P < 0.01 at rest and 8-10 mm, 27-30%, P < 0.05 during MVC). Thickness in GL and SOL increased during MVC compared with rest (by 5-7 mm, 36-47%, P < 0.01 in GL and 6-7 mm, 38-47%, P < 0.01 in SOL) while thickness of GM did not differ (P > 0.05) between rest and MVC. 4. At any given ankle angle the model underestimated changes in GL and SOL occurring in the transition from rest to MVC in pennation angle (by 9-12 deg, 24-38%, P < 0.01 in GL and 9-14 deg, 25-28%, P < 0.01 in SOL) and fibre length (by 6-15 mm, 22-39%, P < 0.01 in GL and 6-8 mm, 23-24%, P < 0.01 in SOL). 5. The findings of the study indicate that the mechanical output of muscle as estimated by the model used may be unrealistic due to errors in estimating the changes in muscle architecture during contraction compared with rest.