Dynamic traffic management (DTM) is the management of traffic streams and of the demand for traffic. Real-time data are essential for correct information to drivers and for control of traffic. Not every variable can be measured directly, so models are necessary to calculate missing information or to predict the future state of the network. This paper follows a top down approach. It starts with the policy context. The different DTM-tools considered are grouped according to their effect either on travel demand, modal split or a specific travel mode. Some recent developments are watched with respect to transport research laboratories. We argue that an emphasis should be laid upon a neat development architecture. For this reason we will pay attention to the traffic model FLEXSYT. The strong point of this model is the use of a traffic control language which allows for the investigation of any type of control for any type of network. Finally we discuss predictability. It is important to identify what can be predicted reasonably and what not.