We studied the rates of revision for 53,698 primary total hip replacements (THRs) in nine different groups of disease. Factors which have previously been shown to be associated with increased risk of revision, such as male gender, young age, or certain types of uncemented prosthesis, showed important differences between the diagnostic groups. Without adjustment for these factors we observed an increased risk of revision in patients with paediatric hip diseases and in a small heterogeneous 'other' group, compared with patients with primary osteoarthritis. Most differences were reduced or disappeared when an adjustment for the prognostic factors was made. After adjustment, an increased relative risk (RR) of revision compared with primary osteoarthritis was seen in hips with complications after fracture of the femoral neck (RR = 1.3, p = 0.0005), in hips with congenital dislocation (RR = 1.3, p = 0.03), and in the heterogenous 'other' group. The analyses were also undertaken in a more homogenous subgroup of 16,217 patients which had a Charnley prosthesis implanted with high-viscosity cement. The only difference in this group was an increased risk for revision in patients who had undergone THR for complications after fracture of the femoral neck (RR = 1.5, p = 0.0005). THR for diagnoses seen mainly among young patients had a good prognosis, but they had more often received inferior uncemented implants. If a cemented Charnley prosthesis is used, the type of disease leading to THR seems in most cases to have only a minor influence on the survival of the prosthesis.