In this prospective study, the reliability of the morphologic analysis of the continuous Doppler signal was evaluated. Three parameters were considered: the maximum systolic frequency (max A), a resistance parameter (RP) and the degree of spectral broadening (SB). The apparatus used allowed us to construct a color arterial image. Seventy-four subjects were included in this study: 17 volunteers investigated by ultrasound only, and 57 patients investigated by ultrasound and arteriography. Anatomic correlation was ontained in 31 operated cases. The threshold values, as determined in the volunteers, were as follows: 3.5 kHz for max A, 0.86 for RP and 58% for SB. Sensitivity was 94% when max A, SB and cartography were combined to detect tight stenoses. The sensitivity for each of the parameters alone was much weaker (61% for max A and 33% for RP). For all carotid lesions, the sensitivity of detection when all parameters were combined was 68%. Specificity was 77%. The study of the carotid bifurcation by the continuous Doppler effect allowed us to identify tight stenosis with reliability. On the other hand, it was impossible to distinguish between normal arteries and moderate stenosis, or between tight stenosis and complete occlusion.