A new research method is proposed for the medium-voltage (MV) vacuum circuit breaker’s (CB’s) mechanical condition monitoring, which combines the mechanism dynamic features simulation and mechanical condition recognition algorithm based on artificial neural networks (ANNs). This method includes three steps: First, the relations between eigenvalues and mechanical failures of a vacuum circuit breaker (CB) through simulation instead of measurement are obtained. In this paper, the mechanism dynamic features of a vacuum CB in failure are simulated; the simulation results indicate that the parameter that can be monitored—main angle—has different characters for different mechanism failures. Second, the eigenvalues for different failure conditions are described by three parameters. Third, mechanical condition recognition of the MV vacuum CB by an algorithm based on ANN is realized. It is concluded by the work mentioned above, both the known mechanical condition type and the new mechanical condition type of the medium-voltage vacuum CB can be recognized with predetermined reliability.