The invention relates to a method and a device for limiting the risk of faults in a control system, in particular a safety-relevant control system, wherein a preferably intelligent actuator controller (AST), by means of the application of a weighted mean value algorithm, calculates a new control value from the two control values determined by means of diverse redundancy by two independent fault-containment units (FCUs), which new control value, in spite of the occurrence of a fault in one of the two FCUs, causes an object to be controlled by the control system to be guided into a safe state, preferably quickly.