We propose an approach to describe and control complex systems based on fuzzy cognitive map (FCM). A mathematical model of FCMs and a calculation method are described as well as a methodology for constructing and developing them that exploits experts who use fuzzy rules to explain cause and effect among concepts. We apply FCMs in a process control problem and demonstrate their effectiveness. We propose a two-level structure for supervisory control of the process, where the supervisor is modeled as an FCM used for failure detection and decision analysis. There is increasing demand for more autonomous, intelligent systems, and the application of FCMs in control and systems may contribute in developing such systems.