Predictive maintenance of high voltage devices is a major issue for all power providers, due to costs and complexity. This article provides some theoretical and experimental principles which can be useful for all maintenance operations performed on high voltage switching devices and some simple methods applicable for assessing the status of all their contacts. Some of the aspects described here are applicable to other high voltage devices like insulators, surge-arresters etc. An original model based on fuzzy-logic algorithms is performed in order to determine the main elements and issues of the predictive maintenance. All these algorithms are integrated into powerful software tools for all types of high voltage equipment maintenance. It could prove the idea of obtaining a higher service efficiency for existing pieces of equipment, with very low costs.