Doppler ultrasound (DUS) is widely used to diagnose and plan treatments for vascular diseases, but the relationship between complex blood flow dynamics and the observed DUS signal is not completely understood. In this paper, we demonstrate that Doppler ultrasound can be realistically simulated in a real-time manner via the coupling of a known, previously computed velocity field with a simple model of the ultrasound physics. In the present case a 3D computational fluid dynamics (CFD) model of physiologically pulsatile flow a stenosed carotid bifurcation was interrogated using a sample volume of known geometry and power distribution. Velocity vectors at points within the sample volume were interpolated using a fast geometric search algorithm and, using the specified US probe characteristics and orientation, converted into Doppler shifts for subsequent display as a Doppler spectrogram or color DUS image. The important effect of the intrinsic spectral broadening was simulated by convolving the velocity at each point within the sample volume by a triangle function whose width was proportional to velocity. A spherical sample volume with a Gaussian power distribution was found to be adequate for producing realistic Doppler spectrogram in regions of uniform, jet, and recirculation flow. Fewer than 1000 points seeded uniformly within a radius comprising more than 99% of the total power were required, allowing spectra to be generated from high resolution CFD data at 100Hz frame rates on an inexpensive desktop workstation.