Polyethylene acetabular cups retrieved at revision surgery were measured by a shadowgraph technique to determine linear wear, and the values were compared with those obtained from radiographs. There was a close correlation between them, although the radiographic measurements slightly underestimated the true wear. Average linear wear rates for surface-replacement components were much greater than those for conventional prostheses with femoral heads up to 32 mm in diameter. Volumetric wear, calculated using a new formula, was found to be less than previously reported in vivo, and similar in magnitude to the results of experimental wear tests in vitro. The volumetric wear rates were greatest for the surface-replacement components and, for conventional components, were found to increase in a linear manner with component diameter.