This paper concentrates on modeling the vehicle routing to develop an evacuation plan for transit-dependent residents during emergency situation. Planning of transit route in the evacuation is formulated as a vehicle routing problem with time windows (VRPTW). An intelligent algorithm, in which genetic algorithm is embedded with simulated annealing is developed to solve the optimization model. A real evacuation network on which 19 pick-up points and 4 shelters are distributed is used to study the proposed evacuation strategy. The relevant results show the feasibility of the mathematical model as well as the efficiency of the solving algorithm.