This thesis investigates pedestrians' path choices when confronted with a decision point in a novel environment. The two main research foci were how different testing setups of virtual environment (VE) affect pedestrian path choice behaviour, and how the visual characteristics of Shape Complexity (SC) and Visual Content (VC) affect pedestrian path choice behaviour.  Two VE testing setups were tested: the type of VE environment and the sampling method. Both were found to have a medium effect on pedestrian path choices. Photorealistic Print VE and Quick Time Virtual Reality (QTVR) VE had an effect on the choices participants made. Testing in a group setting was also found to produce different results than in individual testing.  SC and VC have a significant role on pedestrian path selection. Overall, we compared behaviours in a VC-rich environment to those in a VC-poor environment, which made evident that VC is an important in formulating our path choice behaviour. Bi-variate correlation and multiple regression analysis showed that a number of SC and VC factors affect path choice behaviour. The significant SC factors were occluded edges, legibility, and links; while significant VC factors were colour, cobblestones and signage, people. Increasing our sensitivity to these factors in the guidelines governing the development and the design of urban pedestrian spaces will increase the success of future pedestrian-oriented spaces.