Although Doppler catheter recordings are used to determine coronary flow velocity, their accuracy in the estimation of volumetric blood flow has not been validated. To address this issue, Doppler-derived coronary flow was measured in a canine model and compared with that obtained by means of an electromagnetic flowmeter. A carotid artery-to-circumflex coronary artery shunt was created in six dogs with tubing that incorporated an inline electromagnetic flow device. The circumflex artery was occlusively cannulated by means of a rigid metal stent of known internal diameter, which was placed 2 cm into the vessel, and flow was measured in the stent region by means of a 3F Doppler catheter. Analysis of Doppler shift signals was performed by means of a zero-crossing detector (ZCD) and an off-line fast-Fourier transformation (FFT) system. Flow derived from peak FFT velocities corresponded closely to electromagnetic flow (slope 1.09, r = 0.93), whereas mean FFT and ZCD velocities underestimated electromagnetic flow (with slopes of 0.47 and 0.46, respectively) despite a close correlation (r = 0.92, 0.94). Thus FFT analysis of the Doppler signal with determination of peak velocity gives the most accurate representation of flow, whereas measurements based on ZCD mean velocities may significantly underestimate coronary flow.