Stereo imaging velocimetry is the quantitative measurement of three- dimensional flow fields using two sensors recording data from different vantage points. The system described in this paper, under development at NASA Lewis Research Center in Cleveland, Ohio, USA, uses two CCD cameras placed perpendicular to one another, laser disk recorders, an image processing substation and a 586-based computer to record data at standard NTSC video rates (30 Hertz) and reduce it offline. The flow itself is marked with seed particles, hence the fluid must be transparent. The velocimeter tracks the motion of the particles, and from these we deduce a multipoint (500 or more), quantitative map of the flow. Conceptually, the software portion of the velocimeter can be divided into distinct modules. These modules are: camera calibration, particle finding (image segmentation) and centroid location, particle overlap decomposition, particle tracking, and stereo matching. We will discuss our approach to each module in this paper, and give our currently achieved speed and accuracy for each where available.