From 1982 to 1987, we randomised prospectively 413 patients requiring primary total hip replacements to receive either a Stanmore or Charnley prosthesis. They were reviewed by an independent observer in an attempt to correlate a number of factors including femoral head size with longevity. There were 213 Stanmore hips and 200 Charnley prostheses. At five to ten years (mean 6.5) 76 patients had died and 16 arthroplasties had required revision. Seven were radiologically loose in asymptomatic patients. There was only one case of deep infection. We found no difference statistically in the clinical outcome or in the revision rate of 4% in the two types of prosthesis. The revision rate was greater for trainees than for senior operating surgeons, and there were recognisable technical errors in seven of the nine Stanmore, and four of the seven Charnley replacements which required revision. Retrospective radiological analysis of a random subset of 51 Charnely and 57 Stanmore femoral components showed no difference in femoral subsidence, but in 14 patients who had had bilateral replacements with one femoral component of each type, there was greater early subsidence of the Stanmore prosthesis. Our results confirm that conventional cemented total hip replacements give acceptable results in a general teaching unit, and we found no evidence of any effect of the size of the femoral head on wear or loosening at five to ten years.