This study describes the reliability of a clinically oriented model for three-dimensional movement analysis of head and trunk movements in children with spastic diplegia. The model consists of five rigid segments (head, thorax, pelvis, shoulder line, spine) and includes a detailed analysis of spinal segmental movements. Within and between session reliability during gait was tested in 10 children with spastic diplegia (6-14yrs). Reliability of discrete parameters was assessed with the intraclass correlation coefficient (ICC) and similarity of thorax and pelvis waveforms with the coefficient of multiple correlation (CMC). Measurement errors were calculated for all parameters (SEM, σ). Results indicated acceptable within and between session reliability of discrete parameters for thorax, pelvis, shoulder line, angle of kyphosis and the majority of the spinal segmental angles, reflected by low SEMs (<4°) and most ICCs>0.60. Within and between session waveform errors were below 4°. CMCs ranged from poor to very good, with highest values for movements in the frontal and transversal planes. The angle of lordosis showed lower between session reliability for several discrete parameters, although waveform errors were still below 5°. Head parameters showed lower overall reliability. The results of this study support the reliability of the proposed model. Head kinematic parameters should be interpreted with caution, due to difficulties in standardization. Accurate palpation of the spinal markers, especially the lumbar spine, is critical and demands thorough training of the assessor.