An understanding of test-retest reliability is important for biomechanists, such as when assessing the longitudinal effect of training or equipment interventions. Our aim was to quantify the test-retest reliability of biomechanical variables measured during short-term maximal cycling. Fourteen track sprint cyclists performed 3 × 4 s seated sprints at 135 rpm on an isokinetic ergometer, repeating the session 7.6 ± 2.5 days later. Joint moments were calculated via inverse dynamics, using pedal forces and limb kinematics. EMG activity was measured for 9 lower limb muscles. Reliability was explored by quantifying systematic and random differences within- and between-session. Within-session reliability was better than between-sessions reliability. The test-retest reliability level was typically moderate to excellent for the biomechanical variables that describe maximal cycling. However, some variables, such as peak knee flexion moment and maximum hip joint power, demonstrated lower reliability, indicating that care needs to be taken when using these variables to evaluate biomechanical changes. Although measurement error (instrumentation error, anatomical marker misplacement, soft tissue artefacts) can explain some of our reliability observations, we speculate that biological variability may also be a contributor to the lower repeatability observed in several variables including ineffective crank force, ankle kinematics and hamstring muscles’ activation patterns.