Huge multi-source data is used in traffic information system research .As a popular indicator, travel time reliability can reflect the time variability and volatility of the transportation system，instead of the shortest path as the priority principle of travel decision-making. The research on how to calculate the index is mature relatively, however the significance and the level of the calculating value have not been researched completely. Based on the historical data , the non-parametric test and fuzzy clustering are used to modeling the threshold of travel time reliability. The proposed method is described the quality of reliability. Under the decision-making scenarios, it is applied for drivers to decide the best departure time and route choice. Finally, the established model is validated by using expressway of Shaanxi in the case study.

CICTP