The authors report on their experience with 2164 US exams of the hip performed over last year on newborns, according to Graf's technique. The babies were divided into 2 groups at first examination: risk/no risk. We considered as risk factors: a positive family history, malposition in utero, abnormalities at birth, and being one of twins. Hips 2a+ and 2a- according to Graf were grouped as "borderline hips" (BLH), extrapolating them from negative and positive cases. BLH have generally shown a tendency to spontaneous normalization, although worsening of the condition was observed in 7.9% of cases. A high incidence of positivity (5.1%) was observed in the no-risk group, versus 9% in the risk group. Moreover, there was low agreement between clinical examination and US findings: the need for US screening is therefore pointed out, not only to promptly detect a decentered hip, but also to prevent the development of juvenile coxarthrosis in case of dysplasia. Premature babies were also studied, and those underweight for their gestational age: the incidence of positive subjects in this group (4.5%) was not higher than that found in the no-risk group (5.1%). We report on the phases of hip normalization as related to stage and therapy. Graf's technique, besides being at present the only reproducible one, represents a method common to all radiologist which allows hip measuring and staging.