The purpose of this study was to assess prospectively the initial results of total knee replacement with the standard spherocentric component and to identify retrospectively the pathogenesis of any unsatisfactory results. As a consequence of this study, the femoral component was modified. Thirty-six knees were reviewed twenty-two to fifty-two months after a standard spherocentric total knee replacement, with a mean follow-up of thirty-five months. Twenty-five (70 per cent) of the knees had significant relief of pain and improved function. Eleven knees were not improved because of confirmed loosening in five, supracondylar fracture in two, infection in one, and suspected loosening in three. Ten of these eleven complications became manifest within the first postoperative year. Alignment in the coronal plane was not consistently precise and hyperextension was a recurrent problem in twelve of the thirty-six knees. A modification in the femoral component was made: the stem was lengthened to 10.0 centimeters and the body was reoriented so that there was 5 degrees of valgus and 15 more degrees of posterior rotation than the original design. This modified prosthesis was used in twenty-five additional knees, with definite improvement in alignment, but the follow-up was insufficient to ascertain the effect of the modification on lossening and fracture.