This project proposal is a (second-year) continuation of work started in Fall 1995. The purpose of the initial work was to study the existence of chaotic dynamics in traffic flow. Two phases of work were begun. First, a simulation based on Cellular Automata (CA) methods was developed. CA models the incremental flow of vehicles as they move through a grid of roadway sections. Each vehicle follows a few simple rules based on the surrounding environment. CA has been shown by other researchers to mimic both the macro- and microlevel dynamics of traffic. We have developed a software tool which allows users to specify the size of the CA model they wish to simulate, and how dense they want the initial system. The simulation then outputs various performance characteristics such as counts, speeds, densities, lane-switching behavior, and platoon lengths. Our studies using the simulation show that periodic behavior can be generated and maintained. In practical terms, this means that a system with automated speed control could lead traffic into extremely predictable and repeatable patterns of behavior. These patterns however are sensitive to small changes in density; so actual control algorithms may need to be more adaptive. The second line of work involves the empirical modeling of actual traffic flow data. A methodology (which had to be invented by the research team) for analyzing linear and nonlinear (chaotic) dynamics has been developed and actual data has been obtained; we are currently in a stage of initial data analysis.