Although the link between sagittal plane motion and exercise intensity has been highlighted, no study assessed if different workloads lead to changes in three-dimensional cycling kinematics. This study compared three-dimensional joint and segment kinematics between competitive and recreational road cyclists across different workloads. Twenty-four road male cyclists (12 competitive and 12 recreational) underwent an incremental workload test to determine aerobic peak power output. In a following session, cyclists performed four trials at sub-maximal workloads (65, 75, 85 and 95% of their aerobic peak power output) at 90 rpm of pedalling cadence. Mean hip adduction, thigh rotation, shank rotation, pelvis inclination (latero-lateral and anterior-posterior), spine inclination and rotation were computed at the power section of the crank cycle (12 o'clock to 6 o'clock crank positions) using three-dimensional kinematics. Greater lateral spine inclination (p < .01, 5-16%, effect sizes = 0.09-0.25) and larger spine rotation (p < .01, 16-29%, effect sizes = 0.31-0.70) were observed for recreational cyclists than competitive cyclists across workload trials. No differences in segment and joint angles were observed from changes in workload with significant individual effects on spine inclination (p < .01). No workload effects were found in segment angles but differences, although small, existed when comparing competitive road to recreational cyclists. When conducting assessment of joint and segment motions, workload between 65 and 95% of individual cyclists' peak power output could be used.