The purpose of this paper is to develop and study algorithms for interaction of Software Agents in group decision making support systems intended for logistic problems of control of transport flows on complex city intersections. The paper regards the problem of modeling and solution of the problem of minimization of transportation costs and idle time of vehicles for the chosen version of the route and schedule at multiple intersections. The authors offer a generalized algorithm for solution of the problem of drawing up an optimum schedule for traffic through multiple city intersections with the use of Software Agents and dynamic computer models of transport flows. It is also supposed that modeling software can detect formation of problematic situations, change the duration of one-type operations on separate processors and interact in search of the best solution. The modeling program is controlling intersections under consideration, while the data collection and analysis program interacts with relational database and makes the optimum processors’ operating schedule with the help of a group decision making support system. The processor is the traffic lights of the intersection, and the operation is handling of a transport flow to an assigned direction and time.