One hundred fifteen patients who underwent total hip replacement for osteonecrosis between June 1972 and April 1990 were divided into 3 groups according to the cause of the disorder: (1) osteonecrosis secondary to alcoholism (21 patients), (2) osteonecrosis secondary to hypersteroidism (29 patients), and (3) idiopathic osteonecrosis (65 patients). To determine the differences in short and long term arthroplasty failure rates, these 3 patient groups were compared with a group of 202 patients who received total hip replacement for osteoarthritis. Statistical analyses were carried out on the following definitions of failure: loosening of the acetabular component, loosening of the femoral component, and revision arthroplasty. Radiolucency and postoperative pain scores also were evaluated. A significant difference in the rate of failure because of loosening of the femoral component was found among the 4 groups. Likewise, a significant difference was found among the 4 groups in all revisions or loosenings or both. However, only the comparison between the idiopathic osteonecrosis and osteoarthritic groups showed a significant difference with survival analysis. Second generation cement technique was as significant as any variable relating to failure. The authors conclude that total hip arthroplasty is an equally viable treatment for the 3 types of osteonecrosis examined in this study; however, failure might be more imminent in studies where larger numbers are needed.