This paper describes the development of a novel planning tool (PEDFLOW) capable of simulating pedestrian movement in a range of environments. Pedestrians are modelled as autonomous agents travelling to a predetermined goal and reacting to obstructions (e.g. bus stops, other agents) according to a set of behavioural 'rules'. In order to ensure this rule set represents veridical behaviour, the rules are being developed according to empirical data. Video recordings are used to help identify and quantify different behaviours (e.g. overtaking, yielding, barging) in response to particular obstructions (e.g. bus stop, approaching person). Next, probabilistic estimates of how these behaviours are related to prevailing environmental conditions (e.g. pedestrian density) and individual characteristics (e.g. age, trip purpose) are derived using a combination of observational and interview techniques. The results of a study assessing pedestrian movement in several city-centre locations, selected for their high levels of pedestrian activity, are discussed with relation to the efficient design of pedestrian facilities. For the covering abstract see ITRD E108404.