OBJECTIVE:To compare overground and treadmill running for differences in the three-dimensional angular kinematics of the lumbo-pelvic-hip complex. DESIGN:A within-subject repeated measures design. BACKGROUND:The treadmill is an attractive research instrument as speed and slope are easily controlled and the required calibration volume is reduced. However, the degree to which treadmill running simulates overground running has not been resolved in the literature to date. METHODS:10 able-bodied subjects ran overground and on a treadmill at a self-selected speed. The treadmill speed was matched to each subjects respective average overground speed. The time-distance and the three-dimensional angular kinematic data were captured using a passive marker based motion analysis system. A set of angular and temporal kinematic parameters were extracted from the data and subjected to statistical analyses. RESULTS:Significant differences were found between overground and treadmill running for all the time-distance parameters. Despite this, the kinematics of the lumbar spine and pelvis were similar between the two running conditions, with only three parameters being significantly different. These were lumbar extension at initial contact, anterior pelvic tilt at initial contact and the first maximum anterior pelvic tilt. Hip flexion-extension parameters were also only found to display subtle differences. Of the 17 hip parameters analysed, only hip flexion at initial contact, maximum hip flexion at loading response, hip extension at toe off, maximum hip extension and hip flexion-extension range of motion were found to be significantly different. CONCLUSION:A high powered treadmill with a minimal belt speed fluctuation is capable of being used to obtain a representation of the typical three-dimensional kinematic pattern of the lumbo-pelvic-hip complex during running. RELEVANCE:In order for the treadmill to be accepted as a useful research and/or clinical assessment instrument, it must be demonstrated that it does not significantly alter the performance of the evaluated activity. In this respect, a treadmill with minimal intra-stride belt speed variability and similar surface stiffness to the relevant overground condition is likely to be capable of being used to obtain a representation of the typical human running action for well accommodated subjects.