Past and recent investigations of traffic dynamics mostly rest on averaged data taken over a short period of time, e.g. one minute. They make a qualitative distinction between the generally accepted traffic states called "free flow", "synchronised flow" and "Stop-and-Go traffic" possible. However, the essential car-car-interactions on the microscopic level are concealed. With extensive statistical examinations of single-vehicle traffic data presented in this work one gains new insights especially from a microscopic point of view. By means of time series analyses, correlation functions and by use of histogram methods new evaluation methods for driving behaviour are introduced. These quantities show a strong dependency on the present traffic state, the observed interval of density and the environment (e.g., freeway or city traffic). Microscopic features like synchronisation of velocities ranging over a number of vehicles or decreasing time headways smaller than one second noticeably influence macroscopic proper ties of traffic as expressed in fundamental diagrams. Moreover, on the basis of cross correlation functions connecting flow and density some quantities can be defined to discriminate between the several traffic states quantitatively. These empirical results have impacts on modelling and simulation of traffic flow. A modified cellular automaton comprising aspects of anticipation is introduced and discussed in great detail. Metastable states can be formed and all other criteria are also met to indicate a well-operating traffic flow model. A lot of simulations were done in order to analyse single-lane and multi-lane behaviour and to find out the specific problems coming along with event-driven measuremens methods using "inductive loops". Another focus are density waves to estimate their speed and to elucidate the phase separation and the transition between different traffic states.