Outcomes of nonoperative treatment for developmental dysplasia of the hip were compared between two centres. Eighty children in Centre A underwent staged weaning of the Pavlik harness once three consecutive weekly ultrasounds showed Graf Grade I hips, whereas in 48 children in Centre B, the harness was removed immediately. No statistically significant difference was found in the rate of reintervention (repeat harness treatment, closed or open reduction), avascular necrosis or acetabular index between the two methods, although there was a nonsignificant trend towards higher reintervention rate and lower avascular necrosis rate with immediate harness removal. Age at start of treatment significantly impacted upon initial harness success, reinterventions and avascular necrosis.