A method for estimating the locations of muscle origins and insertions from the measurement of surface landmarks was evaluated using two indirect accuracy tests and a three-dimensional computer graphics program. For each of four lower extremity anatomical segments, a least-squares technique was used to map the measured locations of three landmark targets to their anatomically based locations. The residual errors, obtained from the applications of the least squares, supplied the first indirect accuracy test. These residual errors were between 6 and 12 mm for the four anatomical segments when averaged over ten subjects. The second indirect accuracy test was conducted by comparing the predicted locations of end points on two adjacent segments forming a joint. Errors in aligning adjacent end points were between 12 and 29 mm for three anatomical joints when averaged over ten subjects. A three-dimensional computer graphics program was developed by the authors and demonstrated that the static testing techniques alone were insufficient to evaluate the quality of the muscle origin and insertion estimates. Any evaluation of muscle lengths, velocities and lines-of-action from surface landmarks should examine the estimates made from motion data, and should address both the ability of the model to fit the subjects as well as model's ability to represent the geometry of the musculoskeletal system.