This paper presents the topic of context prediction as one possibilityto exploit context histories. It lists some expected benefits ofcontext prediction for certain application areas and discusses theassociated issues in terms of accuracy, fault tolerance, unobtrusiveoperation, user acceptance, problem complexity and privacy. Afteridentifying the challenges in context prediction, a first approachis summarized briefly. This approach, when applied to recorded contexthistories, builds upon three steps of a previously introduced softwarearchitecture: feature extraction, classification and prediction.Open issues remain in the areas of prediction accuracy, dealing withlimited resources, sharing of context information and user studies.