In order to ensure proper operation of a system, a key aspect of Prognostics and Health Management (PHM) is the accurate prediction of the Remaining Useful Lifetime (RUL). Therefore, one research focus in the area of PHM is the development of new prognostics methods and the improvement of existing ones. In general, no single method has the best performance over the entire state space. Instead of using only one preferred algorithm, an intelligent combination of several prognostics methods in an ensemble can improve the prognostic performance. However, the combination of models represents a heuristic, as there is no guarantee that the fusion of prediction techniques will perform better in each state than the individual best algorithm for itself. In this paper, the existing terminology of ensemble methods is briefly summarized as well as a review of ensemble methods used in the field of PHM is given. In addition, an evolved ensemble method for PHM based on the k-nearest combiner is presented. The method considers a distinction between over- and underestimation of the RUL, the increasing significance of a prediction close to end of life and the end of useful prediction. The method is evaluated using a NASA turbofan dataset. In this case study the prediction of the presented ensemble method is outperforming its member algorithm.