Variable message signs (VMS) is one of the Intelligent Transportation Systems (ITS) technologies and widely used to disseminate traffic information under circumstances to assist en-route drivers in making route choice decisions. This paper presents an investigation of drivers’ response to VMS. In order to determine the potential contributing factors, a stated preference (SP) survey was used to collect data through on-site survey, including driver characteristics, trip characteristics, and driver’s perception towards VMS content. The driver’s response to VMS is divided into three groups, namely staying on the original route, hesitating to select the alternative route, and changing to the alternative route. A total of 300 drivers were interviewed by speaking face-to-face in Xi’an China, and 249 valid questionnaires were returned. With the collected data, three models which are the ordered logit model (OLM), the ordered probit model (OPM), and the multinomial logit model (MNL), were respectively developed to predict drivers’ route choice decision making with MVS messages. The result shows that four explanatory variables, including education level, travel frequency, attention and usefulness of VMS information, are found significantly affected drivers’ route choice behavior in three developed models. Additionally, travel time is also a significant factor in MNL. Well-educated drivers who are usually traveling during non-peak hours and frequent travelers are likely to change their routes corresponding to VMS information. From the perspective of the overall prediction accuracy, the proposed MNL is superior to OLM and OPM. Keywords: variable message sign; stated preference survey; ordered logit model; ordered probit model; multinomial logit model

CICTP