Abstract: In order to construct the use intention model of shared autonomous vehicles(SAV) and analyze its impact factors， use timing and use frequency were used to reflect the use intention for SAV. The attitude latent variables were converted into scores using confirmatory factor analyses and they were introduced into the traditional ordered Logit model. The method of modeling SAV use intention by using the ordered Logit models was proposed. Taking personal attributes， commuting characteristics and attitude latent variables as explanatory variables， the SAV use intention models were established. The results showed that the factors such as whether or not to use carpooling like Didi， paying parking fees at the work place and autonomous vehicles(AV) attitude have significant impacts on use timing and use frequency at different prices(1，2 and 3 yuan/km); the groups that use carpooling like Didi， pay parking fees at the work place and support AV tend to use SAV more frequently and earlier in future; the publics expect to use SAV at a price lower than the current price of the carpooling， i.e.1 or 2 yuan/km.

Key words: shared autonomous vehicles(SAV); use intention; ordered Logit model; use frequency; use timing; attitude latent variable