METHODS A three-dimensional kinematic analysis of lumbar spinal movements with an ultrasonic measuring system was used to distinguish patients with chronic low back pain from those without such pain. OBJECTIVE To investigate the effects of chronic low back pain on the three-dimensional movements of the lumbar spine, and to identify variables that would allow discrimination among patients with chronic low back pain and control subjects. BACKGROUND To the authors' knowledge, no previous studies have described or identified altered spinal and pelvic gait kinematics caused by nonspecific chronic low back pain in all anatomic planes. METHODS In this study, 34 participants with chronic low back pain and 22 subjects without such pain were monitored during treadmill gait. Data from the measuring system operating at 30 Hz were low-pass filtered and normalized to a percentage of the gait cycle. RESULTS Cross-correlations showed almost identical patterns of pelvic (S1) and thoracic (T12) movement curves in all anatomic planes between groups. No statistical group differences were detected for either pelvic or thoracic oscillation amplitudes. However, Student's t test showed significantly higher coefficients of variation (P < 0.01) in all anatomic planes of patients with chronic low back pain than in healthy control patients. CONCLUSIONS The phasic patterns and angular spinal displacements of patients with nonspecific low back pain were shown to be within normal limits. However, the patients demonstrated higher degrees of stride-to-stride variability, representing increased fluctuations in dynamic thoracic and pelvic oscillations. These findings, resulting in less than optimal gait patterns, must be considered in the rehabilitation of patients with chronic low back pain.