Patients/materials andmethods:Weperformed a case-control study.Weanalysed 14 childrenwith CP, aged 6–15 years,whowere followed in our tertiary outpatient child neurorehabilitation unit and 15 controls. There were no significant differences in age (CP 11.4±3.6 years, controls 10.6±2.6 years) or gender between cases and controls. In the CP group 9 children were graded GMFCS I, 5 weregradedGMFCS II, 12 childrenhadunilateral and2hadbilateral CP. Two U-shaped and two 8-shaped trial walks per subject were performed during which the accuracy and precision of the footworn devicewasmeasured using an opticalmotion capture system (Vicon, Oxford Metrics) as the reference system. All subjects then performed a continuous 200-meter walk test at their self-selected pace wearing the foot-worn inertial sensors (Physilog III, LMAMEPFL, Switzerland). Limb-related spatio-temporal parameterswere compared between paretic and control limbs while bilateral gait characteristics were compared between CP and control subjects, using nonparametric analyses. Results: Mean accuracy±precision for both groups was 3.4±4.6 cm for stride length, 4.3±4.2 cm/s for stride velocity and 0.5±2.9◦ for initial contact foot pitch angle. For temporal parameters paretic limbs showed longer stance (61.9±2.5% vs 60±0.9%, P=0.006) and shorter swing (38.1±2.5% vs 39.9±0.9%, P=0.006) phases,with an increase in double support in childrenwith CP (24.8±4.7% vs 20.3±1.7%, P=0.001). For spatial parameters stride length (1.07±0.18m vs 1.32±0.14, P<0.001), speed (1.13±0.23m/s vs 1.39±0.11m/s, P<0.001) and peak angular velocity during swing (385±74◦/s vs 450±41◦/s, P<0.001) were decreased in paretic limbs, with significant differences in foot pitch at both heel strike and toe-off (P<0.001). Both maximal heel clearance (22.7±3.1 cm vs 25.6±3.5 cm, P=0.004) and maximal toe clearance (7.6±2.9 cm vs 13.4±1.6 cm, P<0.001) were lower in paretic limbs. Discussion and conclusions: Foot-worn inertial sensors allowed us to analyze gait kinematics outside a laboratory environment with a good accuracy and precision. The case control comparison yielded results that were congruent with what is known of gait variations in children with cerebral palsy who walk independently. Participants found the system lightweight and easy to wear and use. While not substituting for complete 3D gait analysis, portable sensors provide precise information about gait in conditions that are closer to the child’s habitual environment and motor behaviour, and could therefore prove to be a useful complement.