This paper proposes a hybrid multiagent learning algorithm for solving the dynamic simulation-based bilevel network design problem. The objective is to determine the optimal frequency of a multimodal transit network, which minimizes total users' travel cost and operation cost of transit lines. The problem is formulated as a bilevel programming problem with equilibrium constraints describing non-cooperative Nash equilibrium in a dynamic simulation-based transit assignment context. A hybrid algorithm combing the cross entropy multiagent learning algorithm and Hooke-Jeeves algorithm is proposed. Computational results are provided on the Sioux Falls network to illustrate the performance of the proposed algorithm. assignment problem to obtain a user-optimal equilibrium flow by Frank-Wolfe algorithm and then applied Hooke-Jeeves algorithm to iteratively derive optimal frequencies in a static transit network. Other solution techniques for the bilevel programming problem can be found in [6]. However, for dynamic simulation-based transit assignment, the above derivative-based methods cannot be applied since the functional form of the derivatives is generally unavailable. The simulation-based VI problem is generally difficult to solve in the dynamic transit system. For this issue, Ma and Lebacque [7][8] proposed a cross entropy (CE) based solution algorithm to iteratively derive optimal travel choice probabilities towards user equilibrium based on minimizing the Kullback-Liebler relative entropy (cross entropy) between two consecutive probability distributions.In this work, a hybrid algorithm is proposed by combing the multiagent cross entropy learning algorithm and the Hooke-Jeeves algorithm for solving the simulation-based transit network design problem. The proposed algorithm is derivative-free, convenient for solving the simulation-based TNDP. For the transit system simulation, a multiagent approach is proposed to capture explicitly the transit system dynamics. We propose a multi-layer network to effectively represent the transit network and simulate the movement of different agents (passengers and vehicles). Passenger's waiting time at stop is explicitly calculated subject to the capacity constraint of the vehicle.The rest of the paper is organized as follows. Section 2 describes the mathematical formulation of the bilevel programming problem for the TNDP. It follows in Section 3 the dynamic transit system description based on the multiagent approach along with the transit network model and travel cost formulation. Section 4 presents the proposed solution algorithm by combining the Hooke-Jeeves algorithm and the CE multiagent approach. A state-of-the-art algorithm based on the method of successive average (MSA) for solving simulation-based dynamic traffic assignment problem is proposed. Section 5 provides the computational results of the CE multagent approach and the MSA appraoch on the Sioux Falls network [2] to validate the obtained lower-level user euilibrium solution. Then we show the optimal transit frequency obt...