Travel data of urban residents is an important base for urban planning,traffic management and behavior research.Traditional methods such as face-to-face interview and paper-based questionnaire are of high cost and low accura-cy.This study develops a travel survey system based on GPS-enabled smartphones.A rule-based algorithm is proposed to extract travel information automatically from high-frequency GPS data collected by smartphones.A case study is carried out at New Tongji Village,a residential area in Shanghai.A comparison of travel data obtained from traditional question-naire-based surveys and smartphone-based surveys is also made.The results from the questionnaire survey are used to im-prove the parameters of the trip generation model based on the data from the smartphone-based survey,and the inaccura-cy in traditional questionnaire-based survey is also studied.It is shown that the overall error of the traditional survey is a-bout 33%,and the error for non-home-based travel is as high as 1 59%.Finally,travel behavior of urban residents is also analyzed based on the data from smartphone-based survey.