We have compared different types of intertrochanteric osteotomy for avascular necrosis of the hip and evaluated their performance in the light of improving outcome after total hip arthroplasty (THA). During a period of 14 years we performed 63 flexion osteotomies (partly combined with varus or valgus displacement), 29 rotational osteotomies, 13 varus osteotomies, eight medialising osteotomies and two extension osteotomies. The mean period of follow-up for all 115 operations was 7.3 years (maximum 24.6). At follow-up, 27 of 29 patients with a rotational osteotomy had already undergone a THA, compared with 36 of 63 after flexion osteotomy. A high incidence of complications (55.2%) was seen early after rotational osteotomy, compared with 17.5% after flexion osteotomy. For all osteotomies there was a high correlation between the size of the necrotic area and the incidence of failure, which also correlated with the preoperative Ficat and Steinberg stages. Using Kaplan-Meier survivorship analysis, Sugioka's rotational osteotomy showed a survival probability after five years of 0.26 (95% confidence interval 0.49 to 0.14), and after ten years of 0.15 (CI 0.36 to 0.06). The survival probability for flexion osteotomy was 0.70 (CI 0.83 to 0.59) after five years and 0.50 (CI 0.65 to 0.38) after ten years. The subgroup of flexion osteotomy with a necrotic sector of less than 180 degrees achieved the best survival probability of 0.90 (CI 1.00 to 0.80) after five years and 0.61 (CI 0.84 to 0.45) after ten years. The indications for intertrochanteric osteotomy for avascular necrosis of the hip have to be addressed critically. Even flexion osteotomy in cases with small areas of necrosis provides only temporary benefit. Rotational osteotomy was associated with a high incidence of complications.