This paper presents an experiment on judgments of the perceived complexity of corridor layouts, experienced within a CAVE™-based virtual reality. This work built upon an earlier experiment (Hölscher and Dalton, 2008) in which identical stimuli(the layouts of corridor systems in buildings) were presented in two modes: plan views and movies of simulated walkthroughs. By reproducing the earlier study data has now been gathered for three modes of presentation, the final being a virtual environment. After an initial training level, six randomly ordered stimuli were presented to 20 subjects: ‘experts’ (architects or students currently enrolled on an architectural course) and ‘lay people’ (all others). In each of these corridor systems, subjects were instructed to walk from the starting location (‘A’) to a distant goal (‘B’), and to return unaided. Pointing tasks were undertaken at both locations A and B. They were then asked to judge the complexity of the layout and estimate the difficulty or ease of navigating in that environment. Finally they were asked to identify the correct corridor-layout from a series of similar plans. The aims were to investigate whether there were differences between these two groups in terms of their judgments of building complexity, effects of modality of stimuli (as compared to the earlier experiments) and if any environmental measures (for example information theoreticbased) correlated with the assessments of the two groups. The results were, first, that there is a significant effect of modality, and that the CAVE™ appeared to produce far better correlations with environmental measures of complexity (compared to the plan and movie views) and, second, that the differences between the judgments of the experts and non-experts were far less pronounced in this study, compared to the earlier study, possibly due to the limited number of layout-stimuli or to the relative simplicity of the corridor systems.