Skeletal muscle bulges when it contracts. These 3-dimensional (3D) dynamic shape changes play an important role in muscle performance by altering the range of fascicle velocities over which a muscle operates. However traditional muscle models are 1D and cannot fully explain in vivo shape changes. In this study we compared medial gastrocnemius behaviour during human cycling (fascicle length changes and rotations) predicted by a traditional 1D Hill-type model and by models that incorporate 2D and 3D geometric constraints to in vivo measurements from B-mode ultrasound during a range of mechanical conditions ranging from 14-44 N m and 80-140 rpm. We found that a 1D model predicted fascicle lengths and pennation angles similar to a 2D model that allowed the aponeurosis to stretch, and to a 3D model that allowed for aponeurosis stretch and variable shape changes to occur. This suggests that if the intent of a model is to predict fascicle behaviour alone, then the traditional 1D Hill-type model may be sufficient. Yet, we also caution that 1D models are limited in their ability to infer the mechanisms by which shape changes influence muscle mechanics. To elucidate the mechanisms governing muscle shape change, future efforts should aim to develop imaging techniques able to characterize whole muscle 3D geometry in vivo during active contractions.