Multivariate survival analysis with use of the Cox proportional-hazards model was applied to a consecutive series of 293 primary Charnley total hip arthroplasties performed on 246 patients. The purpose of the analysis was to identify risk factors for, and to quantitate their effects on, aseptic failure of fixation. The duration of follow-up ranged from one month to twenty-three years (average, thirteen years). The end point of survival was defined as radiographic evidence of failure of fixation or as a revision operation.Failure of fixation of the acetabular component was defined as complete demarcation or migration. Failure of the femoral component was defined as progression of at least one of five postoperative signs (subsidence, demarcation of the cement, separation of the component from the cement, fracture of the cement, and endosteal cavitation) or as the occurrence of at least two of these signs. Twenty-four specific items of data for each acetabular component and thirty specific items for each femoral component formed the sets of variables for the analysis.With use of radiographic evidence of failure as the end point, the sixteen-year rates of survival (with 95 per cent confidence interval) were 83.6 ± 5.6 per cent for the acetabular components and 90.9 ± 4.1 per cent for the femoral components. With use of revision as the end point, the sixteen-year rates of survival were 92.3 ± 4.0 per cent and 95.6 ± 3.2 per cent, respectively. The most important risk factor affecting radiographic loosening of the acetabular component was rapid wear of the polyethylene (0.2 millimeter or more annually), followed by the classification of the osteoarthrosis (as hypertrophic, normotrophic, or atrophic) according to the extent of osteophyte formation. The sockets in the hips that had hypertrophic osteoarthrosis survived longer than those in the other two groups. Survival of the acetabular component as determined on the basis of revision was affected only by rapid wear of the polyethylene. Survival of the femoral component, with either radiographic failure of fixation or revision as the end point, was affected by an unfavorable geometry of the medullary canal (a so-called stovepipe canal or a large canal).Patients who have rapid wear of the polyethylene, little osteophyte formation, or an unfavorable geometry of the canal should be followed carefully. These risk factors warrant additional evaluation.