The reliability of the Caterall and the Herring classifications (inter-observer study) was compared using radiographs of 63 hips in 58 patients. The radiographs were reviewed by three independent observers with various clinical experience. The results were analyzed using kappa statistics and the percentage of inter-observer agreement. The less experienced observers seemed to be more prone to classify the hips as more affected than they actually were. They also seemed to produce a higher agreement rate. Disagreement was most common in the Caterall groups II and III. The Herring classification appears to be more reliable, also in children older than six, which perhaps makes it more useful in deciding between conservative or operative treatment.