This paper considers a Safeguard which upon demand should initiate some action within a short time. A form of dynamic redundancy is introduced and the reliability of a Safeguard is analysed on the basis of two computer algorithms. Both fail-to-safe and fail-to-danger probabilities are considered. The advantage of this form of redundancy over conventional r-out-of-n:G systems is discussed, and the algorithms are illustrated by way of examples. Comparative results for various redundant configurations are given.