A comparative statistical analysis was performed weighing the relative merits of two surgical approaches for total hip arthroplasty--the Watson-Jones anterolateral approach and the gluteus maximus splitting posterolateral approach. Intraoperative and clinical records were evaluated for 175 patients undergoing total hip arthroplasty between 1975 and 1979--100 by a posterolateral approach and 75 by an anterolateral approach. Minimum follow-up period was two years. Mean operative times for the posterolateral and anterolateral approaches were 62 minutes and 140 minutes, respectively. The posterolateral group had a mean blood loss of 433 ml, with 35% requiring an average transfusion of 2.5 units. The anterolateral group had a mean blood loss of 767 ml, with 78% requiring an average transfusion of 2.5 units. The average lengths of hospitalization for posterolateral and anterolateral groups were 13 and 15 days, respectively. The time to independent cane ambulation was significantly less in the posterolateral group. The rate of post-operative complications was similar in each group. However, in the posterolateral group there were four dislocations, three loosened femoral components, one deep-wound infection, and one pulmonary embolus, while the anterolateral group suffered one dislocation, no prosthesis loosening, no deep-wound infections, and three pulmonary emboli. Although the posterolateral approach was associated with a lower perioperative morbidity than the anterolateral approach, the latter exhibited fewer dislocations and loosened acetabular components.