Functional specifications for an automated capacitor bank predictive maintenance system that works with a conventional power quality monitoring system, are described in this paper. This system uses raw power quality monitoring data to automatically evaluate and characterize transient disturbances and system conditions associated with the operation of capacitor banks, i.e., during energization, de-energization, and when the capacitor bank is in service. The system is composed of six rule-based modules. The main characteristics and requirements of each module are described. An example output of this predictive maintenance system is provided in order to illustrate the utility and adaptability of the system. A brief discussion of the future system development plans concludes the paper.