Purpose. The goal of this study was to determine whether non immersive interactive virtual environments are an effective medium for training individuals who suffer from Unilateral Spatial Neglect (USN) as a result of a right hemisphere stroke, and to compare it to a standard computer visual scanning training. Method. Participants included 19 patients with right hemisphere stroke in two groups, 11 in an experimental group were given computer desktop-based Virtual Reality (VR) street crossing training and 8 in a control group who were given computer based visual scanning tasks, both for a total of twelve sessions, 9 hours total, over four weeks. Measures included: 1. Standard USN assessments, paper and pencil and ADL checklist; 2. Test on the VR street program; and 3. Actual street crossing videotaped. Testing was performed pre and post intervention. Results. The VR group achieved on the USN measures results that equaled those achieved by the control group treated with conventional visual scanning tasks. They improved more on the VR test and they did better on some measures of the real street crossing. Conclusions. Despite several limitations in this study the present results support the effectiveness of the VR street program in the treatment of participants with USN, and further development of the program.