Three hundred six neonates referred to a research clinic for vibration arthrometry were studied to determine whether the application of objective methods could enhance the accuracy of hip examinations. A total of 629 vibration episodes (hip ``clicks'' and ``clunks'') were detected using miniature accelerometers and recorded during testing for congenital dislocation of the hip. Five categories of congenital dislocations of the hip (CDH) were defined: normal (53 cases), safe click (201 cases), click associated with later CDH (16 cases), unstable (14 cases), and late CDH (22 cases). Follow-up examinations were performed over a four-year period. Detailed measurement of vibration signals was possible and resulted in the resolution of eight variables for each episode. Two variables, pulse area and peak frequency, were selected for their power to discriminate between the five clinical groups. The pulse area values were higher for the late and unstable groups. In addition, the vibrations were examined using Fourier analysis, and the safe click group demonstrated higher frequency components than any other group. Vibration arthrometry is proposed as a possible adjunct in screening for CDH, as it may pick up cases that would be missed by an average examiner and it could monitor clicks, thus avoiding nonessential splintage.