In recent years, the massive public transportation data has become a comprehensive and reliable data source for studying the characteristics of bus travel behavior. Based on the smart card data, GPS data and location information of bus stations in Xi'an, this paper first pre-processing and analyzes these data and deduces the boarding sites using data fusion technology, and then, establishes a judgment model for alighting sites according to the structure and features of the passengers’ trip chain. After getting the boarding and alighting sites, the rule of passenger flow changing with time is analyzed, and the spatial distribution of passenger flow is explored by constructing OD matrix and obtaining desire line, which takes high-tech zone as the main research region. It’s very important to the traffic operation management, which can also provide data support for the optimal adjustment of bus lines and the daily scheduling of buses as well as bus travel organization. KEYWORDS: Massive bus data; data mining; bus on-off stop judging; desire line; travel behavior characteristics

CICTP