In the existing model of travel mode choice behaviors, only the observable characteristics of the traveler's socioeconomic status and travel plan are considered, while the latent variable (LV) affecting the travel mode choice is ignored. In order to solve this problem, the concept of LV in travel behaviors is proposed, and the causal rela-tionships between the LV and the manifest variable as well as between the LV and its measurement variables are de-scribed by using a structural equation model (SEM). Then, based on the maximum utility theory, a SEM-Logit in-tegration model containing both the LV and the manifest variable is constructed by improving the utility function of travel mode in the Logit model. The results show that, as compared with the traditional Logit model, SEM-Logit in-tegration model improves the goodness and the maximum likelihood value respectively by 0. 201 and 20. 607, which means that the LV plays an important role in travel mode choice behaviors and that the proposed model is of higher explanatory ability and accuracy.