Spatial distribution is of particular importance to market analysts, location analysts and transportation planners meanwhile plays an important role in traffic demand management (TDM). For most of people, their living and working places are rather steady, only the places for shopping and recreation can be freely chosen. In this study, an agent-based modeling is applied to simulate destination choices for discretionary activities. Travel diary survey data of Tongling City are utilized throughout the whole process, including extraction of typical activity patterns, formation of agents and reward function, which makes our method more practicable. Thereafter, the accuracy of our model is proved by comparing simulation result to real-world data.