In this paper, generic methodologies for prognostic machinery health management are discussed and a framework is proposed for extending prognostic capabilities to conventional maintenance and diagnostic system (MDS). Two specific techniques that can be used to extract knowledge and rules for fault prediction are discussed. These techniques are based on the analysis of historical data gathered by the MDS. The paper further elaborates on knowledge based real-time failure prediction and recommends a diagnostics-to-prognostics approach that enables traditional reactive MDS to be transformed into proactive health management (PHM) systems.