Objective. One important characteristic in physical activity research into the built environment is network connectivity, usually calculated using street networks. However, a true pedestrian network may have very different connectivity than a street network. This study, conducted in 2004, examines the difference in walkability analyses when street networks versus pedestrian networks are used for four metropolitan suburbs in Perth, Western Australia.Methods. A street network of Perth was used to represent the current standard of data for walkability analyses. Aerial photography from 2003 was used to create a pedestrian network, which incorporated pedestrian footpaths into the street network. The street and pedestrian networks were compared using three measures of connectivity: Pedsheds, link node ratio and pedestrian route directness.Results. A comparison of the results using street versus pedestrian networks showed very different outcomes for conventional neighbourhood designs. Connectivity measures for conventional neighbourhoods improved up to 120% with the addition of pedestrian networks, although traditional neighbourhoods still had slightly better connectivity values overall.Conclusion. The true pedestrian network increases the connectivity of a neighbourhood and may have significant impact on these measures, especially in neighbourhoods with conventional street designs. It is critical that future studies incorporate pedestrian networks into their analyses. (C) 2007 Elsevier Inc. All rights reserved.