The current clinical standard for the analysis of intracoronary Doppler signals is the application of a zero-crossing (ZC) detector. However, the accuracy of the method is questionable, especially in areas of disturbed flow, as confirmed by in vitro studies, animal experiments, and intraoperative observations. The aim of this study is the comparison of a conventional ZC detector and a custom-designed spectral analyzer (fast Fourier transform, FFT) in the analysis of intracoronary Doppler signals obtained in 19 patients undergoing coronary angioplasty. A 3F catheter with an end-mounted Doppler ceramic crystal was placed over an 0.014" guidewire in a normal or near-normal segment proximal to the lesion to be dilated. The Doppler signal was recorded before and after intracoronary infusion of 12.5 mg of papaverine. In 9 patients high flow velocities could be recorded when the catheter was advanced across the stenosis. The blood flow velocity measurements obtained with ZC were significantly lower than the maximal FFT flow velocity measurements (16 +/- 12 cm/s vs. 29 +/- 18 cm/s, p < .001). In all the conditions of Doppler signal acquisition (baseline, hyperemia, stenosis) a large scattering of the signed differences between corresponding measurements was observed. The standard deviation of the difference ZC-FFT was +/- 11 cm/s and +/- 5 cm/s for the maximal and mean FFT flow velocity, corresponding in both cases to +/- 37% of the mean of the ZC and FFT measurements.(ABSTRACT TRUNCATED AT 250 WORDS)