The reproducibility of classifications and measurements used in ultrasound of the infant hip has been evaluated in a prospective, double-blind study. Normal values and range distribution are defined for the main types as well as the potential for discriminating these types by means of ROC-analysis. The clinical validity of the method in defining normal and pathologic findings is discussed in comparison with X-ray diagnosis and follow-up sonographic investigations. Visual analysis focussing on a few diagnostic criteria, mainly the inclination of the acetabular roof and the position of the femoral head, provides sufficiently high accuracy (94% agreement rate) and the best diagnostic results. A correct examination which includes tests of the mobility of the hip joint and limitation of the section level to the maximal depth of the acetabular fossa, allows nearly perfect diagnosing of dysplasia and dislocation.