Public transportation is an important way for the elderly to travel, and walking time to the station is an important part of the quality of public transport services. Taking walking time from the origin to the bus station, and based on the stated preference data, this paper analyzed the impact of walking time on the elderly's bus choice behavior with the random parameters Logit model. Through analysis, it was found that the random parameters Logit model fitting the data much better than the multinomial Logit model. The parameter of in-vehicle time was non-random, while the parameter of walking time was random. Furthermore, the elderly's demographic characteristics including gender, age, motorcar ownership, education, monthly income (greater than 3 000 yuan or not), had a significant impact on the parameter of walking time. So did the latent psychological factors, such as psychological factors towards physical activity, perceived physical functioning. Copyright © 2020 by Science Press.