We made a prospective study of 241 Charnley total hip replacements performed between 1968 and 1974. In 1990, we reviewed 92 patients with 103 hips (96.3% of surviving hips) at a mean follow-up of 17.6 years (15 to 20.6) The clinical results were excellent, with Charnley scores of 4 or more for pain in 95% of the patients, for function in 73% and for movement in 93%. Of the whole series, 8.3% had been revised, and Kaplan-Meier survival analysis showed a probability of revision at 20 years of 10.7%. These results are similar to those from the few other series with extended follow-up, and make it difficult to justify the present widespread use of uncemented hip prostheses. It would seem that some aspects of the design of the first-generation Charnley stem were beneficial for long-term survival of the arthroplasty. We found no correlation between the clinical results and radiological loosening on the Harris scale.