The aim of this work is to determine the effect of this type of treatment on the shape of the femoral head, the range of motion (ROM), radiological changes in the femoral head, and the prognosis of Perthes disease at skeletal maturity. From 1998 to 2007, 53 patients with Perthes disease were treated with a combination of soft tissue release and joint distraction with a hinged monolateral external fixator in 32 patients and by Ilizarov external fixator in 21 patients. Nineteen of our 53 patients attained skeletal maturity and were evaluated in our study. This study included 15 boys and four girls, mean age at surgery 9.3 years (range 7.2–13.1), and mean age at the last follow-up 17.4 years (range 14.9–21.3). The duration of symptoms varied from a period of 6 to 60 months before the operation. Radiographs taken during the fragmentation stage of the disease were classified by the lateral pillar classification of Herring; 19 of our patients attained skeletal maturity and were evaluated. Clinical assessment included the Harris hip score, hip ROM, and limb length discrepancy. Radiographic assessment included sharp transverse acetabular inclination, the uncoverage percentage, the epiphyseal index before surgery (modified Eyre–Brook), at frame removal, and at the last follow-up, the epiphyseal quotient (of Sjovall), and the Stulberg classification. The mean follow-up was 7.2 years (range 4.1–11.3). The mean Harris hip score was 87.1/100 (range 49.2–94.8). An improvement in hip (ROM) of 83.3% of the normal range was restored. There was a marked improvement in the degree of pain and limp postoperatively. The hip ROM was slightly limited in most patients, and seven patients had limb shortening of between 1 and 3 cm. The mean sharp transverse acetabular inclination of the affected side was 44° (range 35–51) compared with 37° for the unaffected side (P=0.042). The mean uncoverage percentage was 36% (range 24–45) compared with 21% for the unaffected side (P=0.027). The mean epiphyseal index was 0.74 (range 0.36–0.94) before surgery, 0.78 (range 0.49–0.89) at frame removal (P=0.017), and 0.80 (range 0.54–0.91) at the last follow-up (P=0.701). The epiphyseal quotient was 0.74 (range 0.51–0.94) and the Stulberg classifications were type II in eight patients, type III in seven patients, type IV in three patients, and type V in one patient. Arthrodiastasis of the hip joint with soft tissue release may represent a good contribution toward the treatment of Legg–Calvé–Perthes disease. This method of treatment has many advantages such as easy technique, minimal rate of complications, a short hospitalization period, correction of shortening because it adds to the length of the limb, and a higher rate of acceptable results than would be expected compared with other methods. It also improves the ROM, reduces superior and lateral subluxation, and provides better radiographic sphericity of the femoral head. In addition, it does not distort the anatomy of the pelvis or the proximal femur; it can be used with equal success in older children who are typically expected to have a poor prognosis. Distraction treatment is not limited by hip stiffness, degree of femoral head deformity, or subluxation, and can be used when other methods of treatment are contraindicated.