Sensors built into current smartphones or in modern facilities are becoming ubiquitous and influence our daily lives more and more. Hence, services that proactively adapt their behaviour to the user's needs is no longer a vision of the future. The proactiveness of such services is achieved by using context prediction algorithms. These algorithms are primarily applied to personal context data of a user, for example to user's motion activities or even to complete daily routines in order to predict a user's next context. Due to this fact the paper presents the most significant legal criteria that have to be considered by the prediction approaches while using personal context data. Following, the criteria are used to legally assess the context prediction approaches. Finally, the resulting consequences are discussed.