delays at intersections are viewed as a major source of travel time uncertainty, particularly for commuter travel during peak periods. Therefore, it is meaningful to investigate the route choice behavior of road users under conditions of variation in intersection delay. To model the risk-averse route choice behavior of road users, various travel time uncertainty measures have been suggested, such as the exponential disutility function (5), the travel time budget (6), effec- tive travel time (7 ), and the later-arrival penalty (8). Many reliability- based traffic assignment models also have been proposed. These models can be classified in three main categories with respect to the source of the travel time variability: network supply uncertainty, demand uncertainty, and road users’ perception uncertainty. In the first category, travel time variability arises primarily from sto- chastic capacity degradation (e.g., adverse weather, accident, traffic signal failures, or road maintenance). Lo and Tung proposed a reserve capacity model to measure the degradable network capacity (9). The road users’ route choice criterion is based on the expected path travel time while accounting for the travel time reliability by imposing the chance constraints. Lo et al. further proposed the travel time budget user equilibrium (TTBUE) model in degradable road networks (6). This model assumes that road users with different risk aversions allow different travel time budgets for their travel, and route choice decisions are based on the least travel time budget. In the second category, travel time variability comes from day-to- day demand fluctuation. Shao et al. proposed a demand-driven user equilibrium principle for modeling the route choice behavior of road users in networks with day-to-day demand variation (7 ). The origin–destination (O-D) demands were assumed to be normally distributed, and then a path-based variational inequality model was formulated. Lo et al. extended the TTBUE model to doubly uncer- tain road networks with stochastic demands and link capacities (6, 10). The demand stochasticity arises from infrequent travelers. The O-D demands are assumed to be exogenous variables with normal distribution in the proposed model. In the third category, travel time variability results from uncer- tainties in both the network and road users’ perception. Mirchandani and Soroush first investigated these two sources of uncertainty in traffic assignment problems (11). A generalized traffic equilibrium model was proposed to incorporate both stochastic travel times and variable perceptions into route choice decisions. Considering the effect of adverse weather on network supply and demand, Lam et al. proposed a reliability-based stochastic user equilibrium (RSUE) model under rainfall conditions (12). In their model, the O-D demands are endogenous variables depending on stochastic traffic conditions. The uncertainties of network supply, demand, and road user perception are simultaneously considered. Modeling the Effects of Turn Delay Uncertainties on Route Choice Behavior in Signalized Road Networks Ji-shuang Zhu, William H. K. Lam, Anthony Chen, and Hong K. Lo 1 Traffic flow fluctuation and traffic signal interruption result in turn delay variations at signalized intersections. Such variations can significantly contribute to road users’ travel time uncertainties and affect their route choices in congested urban areas. A reliability-based stochastic user equilibrium (RSUE) model is proposed to investigate the effects of turn delay uncertainties on the route choice problem. In the proposed model, link travel times and turn delays are considered as correlated random variables with covariance relationships. Differences in these variables are mainly due to uncertainties at signalized intersections and day-to-day fluctuations in demand. Given these uncertainties, the concept of effective travel time is adopted to model the route choice behaviors of road users. The RSUE model also considers the perception errors of road users on effective travel times. A path-based solution algorithm is adapted to solve the RSUE problem. Finally, a numerical example illustrates the application of the proposed model and solution algorithm. Sensitivity tests are used to demonstrate the potential use of the proposed model to assess the accident risks at signalized intersections with heavy turning flows. In the literature, traffic assignment problems with intersection delays have been extensively investigated from both theoretical and practical standpoints (1, 2). However, previous studies view delays at inter- sections generally as deterministic variables. Because of vehicular flow fluctuation and traffic signal interruption, intersection delays are inevitably subject to stochastic variations. This issue has attracted attention, particularly for transportation planning and intersection design. Various analytical or empirical models have been developed to measure such delay variations. Fu and Hellinga developed an ana- lytical formula to estimate the variance of overall delay at signalized intersections (3), and Colyar and Rouphail studied the probabilistic distribution of delays on signalized arterials (4). Although much research has been conducted on estimating inter- section delay variations, this issue has yet not been explicitly consid- ered in traffic assignment models. In congested urban areas, stochastic J.-S. Zhu, School of Economics and Management, Beijing University of Aeronautics and Astronautics, Beijing 100083, China. Current affiliation: Department of Civil and Structural Engineering, Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, China. W. H. K. Lam and H. K. Lo, Department of Civil and Environ- mental Engineering, Hong Kong University of Science and Technology, Clearwater Bay, Kowloon, Hong Kong, China. A. Chen, Department of Civil and Environmental Engineering, Utah State University, Logan, UT 84322-4110. Corresponding author: J.-S. Zhu, zhujishuang2004@163.com. Transportation Research Record: Journal of the Transportation Research Board, No. 2091, Transportation Research Board of the National Academies, Washington, D.C., 2009, pp. 1–11. DOI: 10.3141/2091-01