In order to study the influence of autonomous vehicles on short and medium distance mode choice for intercity travel, this paper incorporates the latent psychological variables that affect the choice behavior of autonomous vehicles into the latent class conditional logit model and establishes a hybrid choice model to conduct the empirical research based on the theory of planned behavior. The results show that compared with the traditional multinomial logit model, the latent class conditional logit model has higher fitting goodness. Travelers can be divided into three subgroups: class1, class2, and class 3, accounting for 40.7%, 24.4%, and 34.9%, respectively. At a 5% confidence level, gender, education, occupation, monthly household income, children, and IC card significantly affect the sample’s latent class. The value of in-vehicle time of class1 and class3 are 2.400 yuan/min and 2.169 yuan/min, which is slightly higher than the total in-vehicle time value of the sample (1.799 yuan/min); the access and egress and waiting time value of class1 is as low as 0.702 yuan/min, while the value of class3 is 8.607 yuan/min. Travelers from class 1 are more sensitive to the travel costs of autonomous vehicles and intercity buses than travelers from class 3 based on the elasticity analysis.