The paper describes the design of a hybrid modelling and simulation methodology, its implementation in the G2 real-time expert system, and its application to a rolling mill. The methodology is specifically designed for industrial control systems in which continuous-time subsystems interact with sequential control stations and the heuristic knowledge of a process operator. The pivotal concept in the methodology is a hybrid model base which is a repository of several paradigms. Hybrid simulation is based on the concurrent execution of the G2 continuous simulator, the G2 inference engine and a Grafcet simulator. The application of the methodology to a rolling mill demonstrates its potential for integrating a hybrid simulation model with the functionality of a real-time process.