One thousand and forty-two McKee-Farrar prostheses of the present design inserted in Norwich from January 1965 to December 1972 have been reviewed retrospectively to determine the incidence of complications needing revision. Of prostheses implanted for more than two years, 6.6 per cent needed revision for loosening (cup 35 per cent; stem 2.2 per cent; both components 0.9 per cent). Of the total number, 2.3 per cent became infected and 1.9 per cent dislocated. Most dislocations needed only a single closed reduction but 08 per cent were revised. The outcome of revision operations was also assessed. Of revisions for loosening, 40 per cent needed no further operation but 23 per cent required excision; pelvic fracture or bone destruction around the components made success unlikely. Revisions for dislocation were disappointing. Of all revisions 17 per cent became infected. Excision arthroplasty is better than a series of failed revisions in an elderly patient.