Virtual courses often separate teacher and student physically from one another, resulting in less direct feedback. The evaluation of virtual courses and other computer-supported educational systems is therefore of major importance in order to monitor student progress, guarantee the quality of the course and enhance the learning experience for the student. We present a technique for the usage evaluation of Web-based educational systems focussing on behavioural analysis, which is based on Web mining technologies. Sequential patterns are extracted from Web access logs and compared to expected behaviour.