Due to physical fitness and travel time different from commuters, the elderly usually travel by bus or on foot. Using the confirmatory factor analysis and binary Logit model, studying the influences of the elderly′s personal statistical attributes and psychological factors on the travel behavior, such as their attitude towards daily travel, their feelings about taking buses at peak times, and their travel intention to avoid rush hours before and after the preferential policy of suspending public transport at morning peak hours, the latent classification model was used to classify and analyze the elderly groups. The results show that all kinds of psychological factors have significant influence on elderly′s short-distance trips. The bus preferential policy and the group′s support for off-peak travel have a significant impact on the elderly′s choice. The latent classification model shows that the survey objects have necessary travel purpose, and the proportion of elderly groups who are sensitive to bus travel and off-peak travel is low.