Two experiments collectively explored four factors that may influence people's judgments of exocentric (interobject) distances in virtual environments. Participants freely navigated in a simple virtual environment and repeatedly made magnitude estimations of exocentric distances. Distances were generally overestimated. An exponential model (Stevens' power law) fit the data, and exponent estimates were generally less than unity. Geometric field of view (GFOV) and the presence of error-corrective feedback were found to have the strongest effect on accuracy. In fact, distance perception was nearly veridical when made with an 80 deg. GFOV and when receiving feedback. Display type (head-mounted versus desktop) and the presence of additional perspective cues were less influential.