Between June 1988 and December 1997, we treated 332 babies with 546 dysplastic hips in a Pavlik harness for primary developmental dysplasia of the hip as detected by the selective screening programme in Southampton. Each was managed by a strict protocol including ultrasonic monitoring of treatment in the harness. The group was prospectively studied during a mean period of 6.5 +/- 2.7 years with follow-up of 89.9%. The acetabular index (AI) and centre-edge angle of Wiberg (CEA) were measured on annual radiographs to determine the development of the hip after treatment and were compared with published normal values. The harness failed to reduce 18 hips in 16 patients (15.2% of dislocations, 3.3% of DDH). These required surgical treatment. The development of those hips which were successfully treated in the harness showed no significant difference from the normal values of the AI for the left hips of girls after 18 months of age. Of those dysplastic hips which were successfully reduced in the harness, 2.4% showed persistent significant late dysplasia (CEA <20 degrees) and 0.2% persistent severe late dysplasia (CEA <15 degrees). All could be identified by an abnormal CEA (<20 degrees) at five years of age, and many from the progression of the AI by 18 months. Dysplasia was considered to be sufficient to require innominate osteotomy in five (0.9%). Avascular necrosis was noted in 1% of hips treated in the harness. We conclude that, using our protocol, successful initial treatment of DDH with the Pavlik harness appears to restore the natural development of the hip to normal. We suggest that regular radiological surveillance up to five years of age is a safe and effective practice.