The principle of fault detection is expanded to fault compensation. The main idea is to use the residual signals generated by a fault detection and isolation system (FDI) in order to reject the effects of faults on the plant. This has the advantage that nominal operation of the plant can be sustained even during the occurrence of a fault. Especially, in potentially dangerous situations this provides time to take countermeasures without letting the system move into critical operation points. The fault compensation algorithm presented in this work superimposes suitable additional signals to the nominal inputs to the plant in order to eliminate the effects of certain faults. The fault rejection dynamics can be designed independently from any existing controller.