the assumption of the availability of historical information on the demand and its distribution in the network. For example, the pioneer work of Cascetta et al. (4) provides a generalized framework for the dynamic O-D estimation problem. The framework provides a least- squares error formulation for estimating the demand in a general network by making use of the modeling results in the field of DTA. The notation of link flow proportion is introduced and describes the fraction of O-D flow that contributes to the flow on a link in a time interval. Xu and Chan (5) estimated the O-D flows without a priori information by introducing dynamic screen lines and assuming travel times from the different O-D pair to these screen lines are known. Dixon and Rilett (9) seem to be the first to incorporate automatic vehi- cle identification (AVI) data to estimate the O-D demand flows. The AVI data are used to extract information on the time-varying link flows, the link flow proportions, and the observed O-D flows. Ashok and Ben-Akiva (6 ) and Kang (7 ) adopted the Kalman filtering technique to estimate the O-D demand flows. While Ashok and Ben-Akiva (6 ) estimated the O-D flow values directly, Kang (7 ) defined the demand variation by using a third-degree polynomial and used Kalman filtering to estimate the coefficients of this poly- nomial. Sherali and Park (11) presented a least-squares formulation to determine the time-dependent trip tables. A column generation approach that uses a sequence of dynamic shortest-path subproblems is adopted for the solution algorithm. Tavana (12) extended the models from Cascetta et al. (4) to consider the consistency between the link flow proportion and the estimated demand. A bilevel optimiza- tion framework is considered. The upper-level problem minimizes the deviation of the estimated link flows from the time-varying link flow observations. The lower-level problem solves for the equilibrium link flows. Zhou and Mahmassani (15) developed a structural-state space model for systematic incorporation of regular demand pattern information, structural deviations, and random fluctuations. The procedure is proposed to capture day-to-day demand evolution and update the a priori regular demand pattern estimate by using new real-time estimates and observations obtained every day. Nonetheless, limited information is reported in the literature on the applicability of these models for real-time traffic management in large-scale urban networks. It is reported that the size of the problem grows considerably with the size of the network (i.e., number of observed links, number of O-D pairs, and estimation horizon) (12). As such, solving the problem for large networks would require inten- sive computation effort that precludes obtaining the estimation results in real time. Unavailability of accurate information on the O-D demand tables is expected to have a significant effect on the robustness of the generated traffic management strategies (1). This paper presents an algorithm for dynamic O-D demand estima- tion that can be implemented in a distributed fashion. Thus, it enables DTA-based traffic management systems to obtain the O-D estima- tion results while meeting their strict real-time requirement. The new Distributed Approach for Estimation of Dynamic Origin–Destination Demand Hamideh Etemadnia and Khaled Abdelghany 127 The problem of dynamic origin–destination (O-D) demand estimation aims at estimating the unknown demand values for all O-D pairs and departure times with the use of available time-varying link flow observa- tions. This paper presents a distributed algorithm for estimating the dynamic O-D tables for urban transportation networks. The new algo- rithm supports the deployment of systems for real-time traffic network management that adopt dynamic traffic assignment methodology for net- work state estimation and prediction. It encapsulates available link infor- mation and reduces the data size required by conventional algorithms for O-D demand estimation. The algorithm adopts a two-stage approach. In the first stage, the study area under consideration is divided into a num- ber of subareas, and an O-D demand table is estimated independently for each subarea. These local O-D tables are then integrated to construct an O-D table for the entire study area. An application of the new algorithm for a typical freeway network is presented as an example. The evolution of advanced traffic management systems over the past two decades has brought considerable attention to the develop- ment of real-time traffic management systems for congested urban networks. The goal of these systems is to provide highway network managers with the capability to develop efficient real-time traffic management strategies to alleviate recurrent and nonrecurrent con- gestion situations. The design of effective strategies depends pri- marily on the availability of accurate information on the spatio- temporal traffic distribution in the network. For instance, development of strategies such as normative route guidance, congestion pricing, and network-based traffic signal control requires information on the dynamic origin–destination (O-D) demand and its distribution along the different routes (1). As the backbone of most proposed traffic management systems, dynamic traffic Assignment (DTA) method- ologies have been used to estimate and predict travelers’ dynamic route choice decisions as function of the evolving network conditions and applied control strategies (2, 3). These DTA methodologies are typically supported by external modules to provide the dynamic O-D demand tables. A primary function integrated into these mod- ules is to estimate the dynamic O-D demand tables through the time-varying link flow observations that are gathered by means of installed surveillance equipment. Several approaches have been proposed in the literature to solve the dynamic O-D estimation problem (4–16 ). The idea is to map the observed time-varying link flows with the demand tables under School of Engineering, Southern Methodist University, P.O. Box 750340, Dallas, TX 75275-0340. Corresponding author: K. Abdelghany, khaled@engr.smu.edu. Transportation Research Record: Journal of the Transportation Research Board, No. 2105, Transportation Research Board of the National Academies, Washington, D.C., 2009, pp. 127–134. DOI: 10.3141/2105-16