Accuracy of space perception and distance estimation in virtual environments is an important precondition for the reliable use of virtual techniques in the design of products, workplaces, architecture, and production systems. The present study compares the accuracy of exocentric 1 distance estimations that a static perceiver achieves with two virtual presentation techniques: a desktop and an immersive cube presentation. Estimation accuracy in a physical mock-up is used as a point of reference. Subjects estimated exocentric distances in detailed models of a workplace previously unknown to them. All distances to be judged were located in the subjects' personal space (less than 1.5 m from the subject). Major differences between the two virtual presentation modes are that stereo information is available in the cube but not in desktop environment, and that, in the cube, changes in perspective are achieved by actually moving inside the cube instead of using a mouse. Furthermore, the cube provides a wider absolute field of view than the desktop environment. The experiment showed advantages of the cube over desktop presentation when estimating exocentric distances in “personal space” from a static position. The magnitude of distance estimation errors was significantly higher in the desktop than in the cube environment. However, estimation errors tended to be overestimations in the cube presentation, whereas over- and underestimation occurred with equal frequency in the desktop environment. In the discussion it is argued that the higher estimation accuracy in the cube environment may mainly be due to the availability of stereoscopic depth cues. According to Cutting (1997), these cues are especially relevant for spatial perception in “personal space.”