Purpose. To evaluate the feasibility and reliability of a novel stiffness assessment tool implemented in the driven gait orthosis Paediatric Lokomat; to investigate the influence of single robotic-assisted gait training (RAGT) on muscle stiffness in children with cerebral palsy (CP). Methods. Ten children with spastic CP conducted a single standard RAGT session and stiffness was assessed before and after the RAGT. Nine of the ten subjects were tested twice on the same day to investigate test-retest reliability, intraclass correlation coefficients (ICCs), standard error of measurement (SEM), coefficient of variation of the method error (CV(ME)) and resistive torques during passive leg movements (stiffness in Nm/ degrees ) were calculated. Results. ICCs showed high reliability (0.83-0.97) for hip and knee movements. SEM and CV(ME) indicated 0.028-0.085 Nm/ degrees , 9.5-23.0% of test-retest variability in hip and 0.018-0.064 Nm/ degrees , 13.3-43.5% in knee measures. Using the assessment tool, a significant decrease in muscle stiffness in participants, especially in children with high levels of muscle tone, could be shown after a single session of RAGT. Conclusions. The assessment tool L-STIFF is a feasible tool for automated measurement of stiffness in children with CP, but it is not sensitive enough to record small changes in muscle tone.