Architects and urban planners design path systems in outdoor and indoor environments, making certain assumptions about how such environments are perceived. We know that people interpret a path system when they do not know the layout of an area, the land uses and activities that lie ahead but specific interpretations are unclear. Preliminary work suggests that pedestrians have preferences although these preferences are not yet codified. This research addresses the question whether the preference is based on simple configuration of the path or whether this configuration is related to interpretations of the layout of the larger environment. In other words, we need to know whether the preference is a fundamental formal preference or cognitively based. The procedure used in this experiment consists of creating an indoor layout with characteristics similar to a nearby existing shopping centre. A virtual environment was created using Bryce modelling software, while surfaces were painted with digital images of shop fronts in the real environment. Several intersection configurations and incident angles were created and the intersections were linked as in a real environment. The path system was then incorporated in QuickTime VR Authoring Studio to allow participants to pan around each choice point before selecting a path. The environment used for conducting this test was the Underground system of Montreal. Storefronts were photographed and used to paint a virtual environment that did not have the configuration properties of the Underground. More specifically, the virtual environment consists of 24 intersections. There are three configurational variables at the intersections: degree of offset, relative width of hall and open space at the intersection. Students were recruited for participation in the virtual navigation exercise. All were told they were entering a virtual shopping centre and were asked to get familiar with it and explore without specific shopping intentions. Twenty- one (21) individuals participated in the experiment in the first stage of the experiment while the sample pool of participants will eventually be increased to around 40. Few of the intersections show statistically robust results due to the relatively small sample size. An examination of the intersections where sufficient numbers of hits were recorded shows two clear trends: 1) a tendency to move straight ahead; 2) a tendency to choose paths where a longer corridor with more tributary corridors could be seen. In the full analysis of the complete participant pool, we will examine the choices by grouping configurational types, a way to strengthen the conclusions with regard to configurational preferences. In general, the statistics used are C2, Wilcoxon-T and power analysis. Finally, we intend to examine the configurational properties of the navigated itineraries themselves. A previous paper using virtual navigation suggested a circular and centering motion when exploring an environment. In general, we are seeing the emergence of three operative factors in virtual exploration navigation: 1) itinerary configuration, 2) information-seeking, 3) configurational preferences. The conclusions of this study will be useful in designing a more focussed study on configurational properties.