Train-based systems are the principal means of public transportation in many of the world's cities, and continue to grow in the face of rising demand. Expanding infrastructure is costly, however, and at a certain point becomes unsustainable. When this occurs the only feasible solution is to improve the management system. This is done by using either offline or online intelligent transportation systems which requires prior analysis and testing. These previous activities are not easy to carry out in the transportation system itself because of high costs and possible drawbacks. The usual solution in these cases involves conducting simulations. Simulating a train system is a complex problem for which several software applications have been designed, using different models, programming approaches, and simplifications. Therefore, selecting the best simulator for testing a particular intelligent system is a hard task that needs atention. In this work, the requirements that a simulator must fulfill in order to be suitable for testing a particular system are stated. For each class of application, examples of available simulators are given and their main characteristics are then analyzed. Finally, as a practical example, the problem of evaluating skip-stop policies in a multi-line Metro system is studied using a novel event-driven simulator.