The purpose of this paper is to combine mode cluster and traveler segment with a integrated structure. Firstly, Nested Logit model was adopted to investigate the alternative similarity. Then a new choice set was redefined by merging those alternatives which were clustered in the same nest. Finally, Mixed logit models were constructed to consider individual heterogeneity. The empirical study focused on the intercity traveling in the western corridor of Taiwan. The result shows that those socioeconomic and trip characteristics (business trip, student, personal income, man group) have significant interactive effects on the both coefficients of travel cost and travel time. The random effect of normal distribution also exists in the coefficient of travel cost. The traveler-segment choice model with reduced choice set has better model fitness than the original choice set. In addtion, this model also obtains more reasonable estimation for value of time.