Background Hip dysplasia is a common cause of secondary osteoarthritis (OA). Periacetabular osteotomy or rotational acetabular osteotomy has been used as jointpreserving procedures. However, only a few reports of long-term results with these operations have been reported. Questions/purposes (1) Would satisfactory clinical and radiographic outcomes be maintained at a mean duration of 20 years after rotational acetabular osteotomy for pre-and early-stage OA; and (2) could we identify risk factors for radiographic progression of OA?Methods Between 1987 and 2001, we treated 159 patients (173 hips) with rotational acetabular osteotomies for the diagnosis of pre-OA or early-stage OA according to the technique of Ninomiya and Tagawa. During that period, our general indications for this approach were age younger than 60 years, center-edge angle less than 20°, and improved femoral head coverage and joint congruency on preoperative AP plain radiographs of the hip in abduction; we did not use this approach when joint congruency was not improved or no widening of the joint space was noted on preoperative AP plain radiographs of the hip in abduction. Of those patients, 21 patients with pre-OA (followup rate: 84% [27 hips]) and 110 patients in the early-stage group (followup rate: 82% [118 hips]) were available at a minimum of 15 years for a total patient group of 131 (followup rate: 82% [145 hips]). The mean age at the time of surgery was 22 years in the pre-OA group and 38 years in the early-stage group. The mean followup was 21 years in the pre-OA group and 20 years in the earlystage group. Clinical evaluation was performed with the Merle d'Aubigne and Postel rating scale, and radiographic analyses included measurements of the center-edge angle, acetabular roof angle, and head lateralization index on preoperative and postoperative AP radiographs of the pelvis. Postoperative joint congruency was also evaluated. The cumulative probabilities of radiographic signs of OA progression were estimated with use of the Kaplan-Meier product-limited method and parametric survivorship analysis using the Cox proportional-hazards model was used to determine risk factors for radiographic OA progression. Results The mean clinical score improved very slightly, from 15 (SD, 0.8) to 18 (SD, 1.1) (95% confidence interval [CI], À2.9 to À2.0; p \ 0.001) in the pre-OA group and from 15 (SD, 1.0) to 16 (SD, 2.1) (95% CI, À2.0 to À1.2; p \ 0.001) in the early-stage group at followup. Fourteen