Gait analysis is actually used in subjects with scoliosis to determine the change in lower limb parameters after surgery, but the reliability of these parameters still remained unknown. The purpose of this study was to assess the repeatability of traditional gait parameters in subjects with adolescent idiopathic scoliosis (AIS) and to estimate the associated standard error of measurement (SEM). A test-retest design was used to assess the reliability of gait parameters at self-selected and fast speeds. A convenience sample of 20 girls aged from 12 to 17 years, with an idiopathic scoliosis (King classification: types I, II or III; Cobb angle 17-50 degrees) participated in the study. Five good trials were recorded on two occasions. The time-distance, kinematic, and kinetic gait parameters were recorded using foot-switches in combination with a three-dimensional motion analysis system (Optotrak) and Advanced Mechanical Technologies Inc., (AMTI) Watertown, MA, USA; force plates. The coefficients of dependability and SEM derived from the generalizability theory were used to assess the reliability. Inter-trial reliability was good for time-distance, kinematic, and kinetic (absolute and normalized) gait parameters except for the medio-lateral ground reaction force (GRF) component and the ankle dorsiflexor moment (phi = 0.60-0.77). Test-retest reliability was higher for the kinetic than for the kinematic parameters. These coefficients ranged from 0.42 to 0.58 for the time-distance parameters; from 0.55 to 0.88 for the angular displacements and from 0.25 to 0.99 for the kinetic parameters. The SEMs were lower than 3.3 degrees for the angular displacements and lower than 8 Nm (0.15 Nm/kg) and 36 W (0.54 W/Kg) for the joint moments and powers regardless of the speed. Several gait parameters are reliable among subjects with AIS and can be used to assess the evolution of the spinal modifications and the impact of treatment on their lower limb gait pattern.