In this study we investigate the effects of computer graphics eyepoint elevation, display type (perspective vs. stereoscopic), and location of computer-generated images within a particular quadrant on the accuracy of spatial judgments. Twelve volunteer participants judged the azimuth and elevation separation between two computer-generated cubes using different computer graphics eyepoint elevation angles and either a stereoscopic or perspective display. The results indicated that eyepoint elevations between 15° and 45° led to the best overall performance for judgments of both elevation and azimuth. Whereas an eyepoint elevation of -15° resulted in significantly degraded judgments in azimuth because of depth compression, an eyepoint elevation of 75° resulted in significantly degraded judgments in elevation because of vertical compression. Furthermore, when objects were located in the quadrant closest to the computer graphics eyepoint, significant degradation in elevation judgments were observed because of vertical compression. In addition, the data indicated that stereoscopic viewing did not enhance performance over perspective displays because the monocular depth cues in the scene were effective in aiding participants in their performance of the spatial tasks. We discuss the implications of the results for display design.