The survivorship of total hip replacements in patients operated on at the Royal National Orthopaedic Hospital has been analysed using standard actuarial methods. Between 1963 and 1972 173 metal-on-metal prostheses of various developmental designs were inserted. Between 1969 and 1972 248 prostheses of one design using a metal femoral component and plastic cup were used. The patients were surveyed retrospectively and for each the actual follow-up period was determined as well as the eventual outcome. The criterion of "survival" was that the prosthesis was still in situ and all such prostheses were regarded as "survivors", regardless of whether the patients experienced pain or loss of function. The survivorship was then determined by constructing life tables. The results indicate that for metal-on-metal prostheses the overall probability of survival was only 53 per cent after 11 years and the average annual probability of removal, irrespective of cause, was 5.5 per cent. The results were better for metal-on-plastic prostheses with figures of 88 per cent after eight years and 1.5 per cent respectively. For both prostheses the predominant reason for failure was loosening and the annual rate of removal for loosening increased as the follow-up time increased, suggesting that loosening was a wearing out process. The advantages of the survivorship method of analysis as compared with conventional methods are discussed.