A computer algorithm has been developed for segmentation and three-dimensional (3D) reconstruction of Doppler color-flow images. The algorithm enables the user to select a range of velocities, represented by colors, for segmentation and subsequent 3D reconstruction. The reconstructed flows are assigned a color palette and merged with the volume-rendered gray-scale image to produce a 3D image containing both flow and anatomic information. The results demonstrate the application of the algorithm to regurgitant and shunt jets with complex spatial and velocity patterns. We conclude that 3D reconstruction of selected color spectra (e.g., velocities) of Doppler color flows and surrounding anatomy is feasible in the clinical setting.