We reviewed the long-term functional and roentgenographic outcomes of severe Legg-Calve-Perthes disease in two similar groups of patients, one of which was managed with a shelf arthroplasty and the other, by a variety of non-operative means. Twenty hips in nineteen patients were treated with traction and, when necessary, adductor tenotomy to attempt reduction of the subluxation; this was followed by a shelf arthroplasty to increase the arc of the acetabulum and result in coverage of the femoral head. The average duration of follow-up was nineteen years (range, two to forty-seven years). The initial center-edge angle and the anteroposterior Mose sphericity measurement were significantly improved at the most recent follow-up (p greater than or equal to 0.05), and the average Iowa hip score was 91 points. Hinge abduction, which was producing incongruity, was eliminated in eleven of fourteen hips after the operation. The outcomes for the second group, consisting of eighteen hips in seventeen patients who were managed non-operatively, were also reviewed, at an average duration of follow-up of twenty-eight years (range, seven to forty-five years). No significant improvement in the center-edge angle or the anteroposterior Mose sphericity measurement was observed at the most recent follow-up, and the average Iowa hip score was 81 points. There was no significant difference in the range of motion of the hip between the groups. We believe that shelf arthroplasty is a favorable option in the treatment of Legg-Calve-Perthes disease when additional bone is needed at the lateral aspect of the acetabulum to cover the extruded portion of the femoral head, to prevent additional incongruity and thereby delay the early onset of degenerative changes.