This paper proposes a classification of system behaviours that allows the generic control methods to be identified and clearly related. It then presents a strategy for integrated process supervision that provides for the controlled evolution through the different control regimes. The generality of the strategy is independent of a particular realisation of individual control regimes, thereby allowing AI methods and 'classical' control techniques to co-exist within an integrated framework. The system has been fully implemented on a SUN 3/160 workstation and validated on a laboratory scale process rig. An overview of the results obtained from a scheme utilising a standard 'Three-Term' controller, a simple adaptive system and fault diagnosis based on a qualitative model of the process rig is presented.