Walking as a non-motorized mode of transport is an essential component of sustainable environment. To increase the share of walking in transportation and accordingly to plan and manage pedestrian areas, a deep understanding of pedestrian movement behaviour is required. The aim of this thesis is to provide walking behaviour models based on discrete choice framework to study the behaviour of pedestrians in different situations. Data has been extracted from real-world video recordings of pedestrian crowds for calibration and validation purposes. In the first part of this thesis, the behavior of pedestrians walking in groups has been modeled based on the tendency of group members to maintain group unity. In the second part, a simulation tool has been developed to calibrate and validate a model presenting pedestrian behaviour at high density bottlenecks. The validation results confirm the acceptable performance of walking models proposed in both parts of the thesis.