Walking, as both a major mode of transport and the most common form of every-day physical activity, deserves further attention in urban geography and public health studies. The control of self-selection has not attracted enough attention in the empirical studies in China, and there is insufficient research on the behavior of distinguishing different walking purposes. Based on the micro survey data of the built environment and residents' travel behavior in Xi'an city in 2014, the quasi-experimental study design of matching case control was adopted. By controlling the self-selection effect brought by travel attitude, this paper explored and distinguished the impact of urban rail transit itself and built environment on walking behavior. Research findings are listed as follows. (1) Self-selection factors such as travel attitude and preference have a critical influence on individual walking behavior. (2) After controlling of individual socio-economic attributes and self-selection factors, the respondents' perception of built environment greatly affected the walking frequency. (3) There is a certain difference of the impact of objective built environment variables on walking behavior between China and developed countries. No significant effect of density was found in walking frequency models of transport or leisure walking. The number of shopping stores and the number of bus stations in the objective built environment variables has a significant positive effect on the transport walking frequency. However, for leisure walking, these built environment variables do not have significant impact. (4) Subway has a significant independent influence on the transport walking frequency. These conclusions can help to further understand the influencing factors and mechanism behind urban residents' pedestrian travel behavior and provide enlightenment for urban transportation and urban planning policies that guide urban residents to green and healthy travel by optimizing urban land use planning.