Since the introduction of transcranial Doppler sonography in the early 1980s, flow velocity estimates have assumed a 0- to 30-degree angle of insonation. Based on limited radiological and anatomical studies, such as assumption appeared justified, and seemed to confer only minimal potential for error due to the cosine function in the Doppler formula. The introduction of transcranial color duplex sonography allows the direct evaluation of this assumption and the effect on flow velocities. Fifteen healthy volunteers were studied bilaterally using a unilateral transtemporal approach from the right. Velocity measurements were taken from the middle, anterior, and posterior cerebral arteries. Flow velocities were obtained with and without angle correction (0 degree). After completion of the color duplex study, velocities were obtained with a conventional, "blind" Doppler transducer at corresponding depths. For all insonated vessels the average angle of insonation was around 30 degrees. However, there was a wide variability of individual angles of insonation (0-70 degrees) in specific vessels. In 74.5% of all vessels, the angle-corrected flow velocity did not exceed the uncorrected velocity by more than 25%. In 14.5% the angle-corrected velocity was 25 to 50% higher and in 10.8% it was more than 50% higher as compared to the uncorrected velocity. Thus, the angle of insonation was unpredictable and often higher than originally expected. Angle-corrected velocities were higher than uncorrected values, and were more than 25% higher in about one-fourth of the vessels studied. Understanding of the clinical importance of such differences requires further study.