The wear of polyethylene acetabular components is an important issue in total hip arthroplasty. The amount of wear has been measured in many studies, but few have addressed other mechanical aspects of wear in vivo. We used the shadowgraph method to measure the wear vectors in 104 retrieved acetabular cups that had been made by a single manufacturer, and we identified more than one wear vector in thirty-one cups (30 per cent). We hypothesized that the most likely explanation of multiple wear vectors was loosening of the acetabular implant with a change in the orientation of the implant in the pelvis. To test this hypothesis, we estimated the extent of motion of the cup in situ on the basis of differences in angles measured from serial radiographs of sixteen hips. We then used linear transformation of the three-dimensional vectors to compare the wear directions measured in the retrieved implants with the wear directions predicted from the radiographs. The change in wear direction predicted on the basis of in vivo motion of the cup never corresponded to the actual difference between wear vectors in the retrieved implants. Our results suggest that multiple wear vectors may be commonly found in retrieved implants, but loosening of the acetabular cup does not account for the multiple vectors. Additional observations suggest that impingement between the edge of the acetabular cup and the femoral component may be associated with multiple wear vectors. These results have implications for the investigation of the general mechanisms of wear in vivo and suggest that clinical or wear-testing scenarios that assume a single direction of wear may underestimate the over-all amount of volumetric wear.