The purpose of this study wa9 to analyse the gait patterns of two female children with hemiplegic cerebral palsy by using the three-dimensional video motion analysis technique. Case 1 has mild spastic hemiplegia on the right side while Case 3 has moderate spastic hemiplegia on the left side. A group of 10, normal female children of the same age(7-8 years old) were selected as the control group for comparison. Time and distance variables as well as the Center of Mass displacement, and the pelvic and joint motions in three anatomical planes were analysed for this purpose. The following observations were made through the analysis : Case 1 revealed an asymmetrical gait pattern in which the step length of the unaffected side was shorter than that of the affected side, which wan a result of the effort to minimize loading on the affected leg by shortening the swing phase of the unaffected leg. Case 1 scored similar phase ratios, cadence and walking velocity to the normal group. A slight posterior tilt of the pelvis was observed throughout the gait cycle. Less hip and knee flexion than the normal group was observed, and demonstrated hyperextension of the knee in the terminal stance phase. The main problem in case 1 originated from the insufficient dorsiflexion of the affected foot during the swing phase. Therefore, Case 1 has difficulty with foot clearance in the swing phase. Usually, this is compensated for by using exessive hip abduction and medial rotation in conjuction with trunk elevation as well as increased vortical displacement of the center of mass. Case 1 revealed a foot-flat initial contact pattern. Case 2 was characterized by a consistent retraction ef the affected aide of the body througout the gait cycle, As a result, an asymmetrical gait pattern with increased stance phase ratios of the unaffected side was observed. In spite of this the step lengths of both sieds were similar. Case 2 scored lower cadence and walking speed than the normal group with lower gait stability. The main problem in Case 2 originated from an excessive plantaflexion of the affected foot which, in turn, rebutted in high hip and knee flexion. Hyperextension of the knee was observed at mid-stance, and execessive anterior tilt of the pelvis throughout the gait cycle was noticed. A gait pattern with high hip abduction and medial circumduction was maintained for the stability in the stance phase and foot clearance in the swing phase. Case 2 revealed a forefoot-contact initial contact pattern.