UNLABELLED: An increasing number of studies address the use of virtual environments (VE) in the cognitive assessment of spatial abilities. However, the differences between learning in a VE and a real environment (RE) remain controversial. PURPOSE: To compare the topographical behavior and spatial representations of patients with traumatic brain injury navigating in a real environment and in a virtual reproduction of this environment. METHODS: Twenty-seven subjects with moderate to severe traumatic brain injury were consecutively included and allocated to one of two groups. The subjects were taught the same route in either the virtual environment or the real environment and had to recall it twice immediately after learning the route and once after a delay. At the end of these sessions, the subjects were asked to complete three representational tests: a map test, a map recognition test recognition and a scene arrangement test. RESULTS: No significant difference was found between the two groups with regards to demographics, severity of brain injury or episodic memory. As a main result, the number of error rates did not significantly differ between the real and virtual environment [F (1, 25) = 0.679; p = 0.4176)]. Scores on the scene arrangement test were higher in the real environment [U = 32.5; p = 0.01]. CONCLUSIONS: Although spatial representations probably differ between the real and virtual environment, virtual reality remains a trusty assessment tool for spatial abilities.