The theory of 'natural movement' postulates that configuration of the urban grid is an important generator of aggregate patterns of movement in urban areas (Hillier et al. 1993). Retail and commercial land uses locate themselves at these configurationally hotspot locations to take advantage of the economic opportunities created by movement i.e. passing customers (Hillier 1996). These retail and commercial areas are also work places for a good number of people and will influence the choice of some residential locations. Since journey-distance and journey-time are two very important factors influencing propensity for walking or cycling (Plaut 2005; Wardman et al. 2007; Pucher & Dijkstra 2003; Schwanen & Mokhtarian 2005) we hypothesize that the sites of retail areas as understood by their configurational index, will first affect the choices of residential locations and hence also influence the use of non-motorized transport (NMT) particularly walking and bicycling. We test this hypothesis in the cities of Pittsburgh and Lubbock (USA) using data collected from American Community Survey and US census bureau. Topological and angular configuration analyses of CAD drawn axial lines and street centrelines derived from GIS maps were performed for both cities. ArcGIS spatial analysis tools were applied to combine land use, socio-economic & demographic, transportation and space syntax variables to the scale of census block-groups that was selected as the study unit. Statistical analyses including stepwise and best-subset regressions were carried out to select relevant and significant variables explaining the use of NMT. The findings indicate that choice of transportation mode is significantly explained by multiple variables in which configuration features prominently. In the paper, we also describe and test our assumptions, comment on the selection of areas instead of axial lines as units of analysis, and postulate on the applicability of this research on creating sustainable communities.