A 2D marker-free method for tracking the human lower limb segments has been developed and tested. The method is based on model matching using distance transformations, and is intended for use in analysis of human gait. The model used consists of five line segments, which are matched to the contours of the lower limb segments. Experiments have been made on two different kinds of objects: a rigid model of a leg and a human leg. Only movement in the sagittal plane has been considered. An image sequence of the moving object was captured, and the method was used for tracking the segments of the object in each image. The results were then compared to data obtained using markers mounted at the ankle joint, the knee joint and the hip joint. The experiments on the rigid body were done to assess the accuracy of the method under perfect conditions. The average difference between the estimated joint positions and the reference positions was found to be less than 2 mm for all three joints, with a standard deviation of less than 1 mm. To test the method in a more realistic situation, it was applied to a human leg during the stance phase of walking. Measurements were made on three healthy male subjects. The average differences between the estimated joint positions and the reference marker positions were less than 6 mm, 10 mm and 22 mm, for the ankle joint, the knee joint, and the hip joint, respectively. The corresponding standard deviations were less than 4 mm, 7 mm and 9 mm, respectively. The satisfying results, though obtained controlled laboratory conditions, show that this method might be useful in gait analysis.