In order to address gaps in the literature, this study assessed the reproducibility (i.e., difference between and within sessions) of joint and muscle forces using wearable sensors during stationary cycling. Seventeen male cyclists performed two sessions on a cycle ergometer cycling at a combination of three power outputs (1.5, 2.5 and 3.5 W/kg) and three pedalling cadences (60, 80 and 100 rpm) in two sessions (2-7 days apart). The first trial from each session was repeated at the end of the session for assessment of within-session reproducibility. Three-dimensional (3D) full-body motion and 3D bilateral pedal forces were collected using an inertial motion tracking system and a pair of instrumented pedals, respectively. Joint angles, muscle forces and knee joint forces were computed using OpenSim. Poor to excellent agreement (ICCs = 0.31-0.99) was observed and differences were trivial to small and non-significant between trials within-session. Poor to excellent agreement (ICCs = 0.05-0.97) was observed and differences were trivial to large between sessions. Variability can be attributed to changes in muscle recruitment strategies (within and between-sessions) and to repositioning of sensors (between-sessions).