Pain following total hip arthroplasty is a common clinical problem, often due to loosening of the prosthetic components. Current methods of detecting loosening and of differentiating it from other causes of pain share certain limitations in sensitivity, specificity and invasiveness. The purpose of this study is to adapt and evaluate a computer assisted stereophotogrammetric technique for clinical application. The technique, originally devised for aerial topography and later adapted for craniofacial measurement, consists of quantitative stereoradiography with computer interface. Metallic beads, implanted during arthroplasty in adjacent bone, provide constant references. A specially designed cage cassette holder with fixed opaque markers permits correction in radiographic geometry. Postoperatively, serial stereoscopic radiographs of the hips are obtained with and without stress. Measurements are made from the resultant radiographs using an xy plotter and an interactive computer terminal. The data points are analyzed and corrected for variation in rotation, magnification, and projection occurring between serial stereo pairs. Thus, abnormal motion between the prosthesis and reference points can readily be computed. Preliminary results on phantoms indicate an accuracy of 0.8 mm in linear measurement at the 95% confidence level. Quantitative data reduction on a selected patient study suggests that measurements comparable to the phantom studies for precision and accuracy are possible. The method may prove to be a sensitive, precise and noninvasive technique for detecting the presence and amount of prosthetic motion in the clinical setting.