This thesis investigates the opportunity to use the massive amounts of data coming from modern aircraft to predict maintenance tasks. The goal is to predict defects in such a fashion as to prevent the actual failure from occurring. This results in fewer delays and less maintenance costs. The aircraft data, ACMS data, is analysed using three anomaly detection modules. The three methods are correlation based, subsequence based and noise based anomaly detection. Results show very promising and have resulted in the preventive removal of 17 components within the KLM B747 fleet to this date.