In this paper a new robust optimization (RO) model is proposed for route guidance based on the advanced traveler information system. The arc travel time is treated as a random variable, and its distribution is estimated from historical data. Traditional stochastic routing models just minimize the expected travel time between the origin and the destination. Such approaches do not account for the fact that travelers often incor- porate travel time variability in their decision making. Recently some RO models were proposed to incorporate more statistical information into the models, but these models still ignore much information available from historical travel time data. The probability measurement, time at risk (TaR), is introduced in this paper, and a multiobjective model is built up that allows a trade-off between the expected travel time and the TaR. Thus, the skewness and kurtosis of the arc travel time distribution are taken into consideration; that is important because the travel time distri- butions of typical arcs show high asymmetry and long tails on the right side as a result of the impact of random incidents and events. This approach is applied in two examples, one of which is a real traffic network. With the advent of the deployment of intelligent transportation systems technologies, especially the advanced traveler information system (ATIS) and advanced traffic management systems (ATMS), the pre- diction of short-term travel time has become increasingly important. The focus of this research is to enhance the route guidance systems, which include information centers and in-vehicle navigation systems. The model will be used by the information centers, which collect and process the travel time information and then send the recommended route and other useful information to different travelers. By collecting information through the telematics and in-vehicle devices equipped in more and more vehicles, traffic information centers can obtain rich, historical traffic information. More accurately estimated future travel time and suggested route choice can be provided to travelers. However, this is not necessarily true if the randomness of the travel time is ignored. For example, travel time in a particular arc will change significantly if there is an incident or a special event, which frequently occurs in everyday life. Even with weather change, the travel time will be different. To provide more reliable route choice suggestions, the bound or the distribution of the travel time should be considered explicitly. Although stochastic routing has been widely used in recent years, most of the current stochastic route choice models are based on min- imizing the expected travel time between the origin and the desti- nation. The best route choice can be calculated easily using deter- ministic shortest travel time models. However, the historical infor- mation is not used efficiently. Travelers cannot incorporate travel time variability into their decision making and they cannot distinguish between cases in which arc travel times are correlated and those in which travel times are uncorrelated. In recent years a body of literature has been developing under the name of robust optimization, in which researchers give a robust solu- tion that takes the data variability into consideration (1). Kouvelis and Yu propose a framework for robust discrete optimization using the minimax approach, which seeks to find a solution that optimizes the worst-case performance under a set of scenarios for the data (2). Ben-Tal and Nemirovski proposed less conservative models by con- sidering uncertain linear problems with ellipsoidal uncertainties, which involve solving the robust counterparts of the nominal problem in the form of conic quadratic problems (3). Bertsimas and Sim present a technique tailored specifically for polyhedral uncertainty that leads to linear robust counterparts while controlling the level of conserv- ativeness of the solution (4). Moreover, their methods readily extend to discrete optimization problems. All three of these approaches are general robust optimization methods to deal with the data uncertainties. However, these existing approaches cannot be applied directly in the stochastic routing cases. To simulate the traveler’s behavior closely, Sen and Pillai, using the idea of a multiobjective model, developed a mean–variance model for route guidance, which can help in this paper to find a trade-off between mean travel time and travel time variability (5). In this research a new robust model is proposed for route guidance; the model considers not only the expected shortest travel time between the origin and the destination, but also a probability measurement proposed here, the time at risk (TaR), which travel time will not exceed under some probability level. The idea of this model comes from the modern theory of portfolio optimization, but it fits very well in route selection problems. In real life when people make route choice deci- sions, they are not concerned only about the travel time on the average, but also about the possible delays in the worst case. As an example, a person may tell a taxi driver that he or she needs to get to the airport as soon as possible and wants to make sure that it will be possible to arrive there before 4:30 p.m. This simple example can help in finding interesting rules if it is analyzed from the view of probability. On the one hand, the aim is to minimize the expected travel time to the air- port. On the other hand, the aim is to find a route to reach the airport before 4:30 p.m., even in the worst case. However, these two objec- tives are not totally consistent with each other. Thus, a trade-off will have to be arranged between them. In the model proposed the objec- tive function is formed as a linear combination of the expected travel time and the TaR, as will be shown in the next section. Meanwhile its performance will be compared with other current robust optimization (RO) models, especially the mean–variance model. The data used Robust Route Guidance Model Based on Advanced Traveler Information Systems Jiangang Lu, Xuegang Ban, Zhijun Qiu, Fan Yang, and Bin Ran J. Lu, X. Ban, Z. Qiu, and B. Ran, Department of Civil and Environmental Engineering, University of Wisconsin–Madison, Madison, WI 53706. F. Yang, ESRI, 380 New York Street, Redlands, CA 92373. 1 Transportation Research Record: Journal of the Transportation Research Board, No. 1935, Transportation Research Board of the National Academies, Washington, D.C., 2005, pp. 1–7.