It would be preferable if drivers could be warned about traffic jams such that they can take another route to their destination. In this thesis, methods for detection of traffic jams are designed. A data warehouse containing historical traffic information is designed to test whether it can be used to detect non-periodic traffic jams.A prototype of the design is implemented to test the three methods on realistic data generated by a data generator. The results from the methods are compared to each other and combinations of the three methods are also considered. The experiments indicate that traffic jams can be detected and the detection process can be improved by optimizing different parameters. The force of data warehouses, namely pre-aggregation, is not used. Therefore, a normal relational model could have been used instead.