Utilities of travelers' arrival modes to a hub station are important factors that affect travelers' mode choice． There is heterogeneity among travelers' preference of mode choice，and the traditional discrete choice mode can't explain the heterogeneity clearly． This study incorporated the heterogeneity into travelers' mode choice to the high-speed railway station with the latent class conditional Logit model． The latent class conditional Logit model produced segment-specific parameter estimates to capture heterogeneous preferences for choice alternatives and identified travelers' profiles for each segment． The results show that the sample population can be divided into three subgroups，and the parameters of in-vehicle time，walking time and price varies in each subgroup． Further- more，the car-ownership，monthly income and latent variables (flexibility，comfort，reliability) have significant influences on the category classification of the sample population．