While feedback-controlled treadmills with a virtual reality could potentially offer advantages for clinical gait analysis and training, the effect of self-paced walking and the virtual environment on the gait pattern of children and different patient groups remains unknown. This study examined the effect of self-paced (SP) versus fixed speed (FS) walking and of walking with and without a virtual reality (VR) in 11 typically developing (TD) children and nine children with cerebral palsy (CP). We found that subjects walked in SP mode with twice as much between-stride walking speed variability (p<. 0.01), fluctuating over multiple strides. There was no main effect of SP on kinematics or kinetics, but small interaction effects between SP and group (TD versus CP) were found for five out of 33 parameters. This suggests that children with CP might need more time to familiarize to SP walking, however, these differences were generally too small to be clinically relevant. The VR environment did not affect the kinematic or kinetic parameters, but walking with VR was rated as more similar to overground walking by both groups (p = 0.02). The results of this study indicate that both SP and FS walking, with and without VR, can be used interchangeably for treadmill-based clinical gait analysis in children with and without CP.