Traditionally, finite automata have no mechanism to represent the time taken between actions. However, in many specifications, we need to model time. For instance, we may want to specify a minimum and maximum delay between turning on a switch and its effect being observed. In 1992, Alur and Dill introduced a model of timed automata in which real-valued clocks are incorporated into the state-transition structure of traditional automata in such a way that some useful questions about the automata remain computable. In this talk, we will introduce the Alur-Dill model of timed automata and look at some technical issues involved in analyzing these models.