Introduction Extensive research has been carried out on the reliability of redundant systems. Several methods and methodologies have been discussed in monographs by Birolini[1] and Srinivasan and Subramanian[2]. Liebiwitz[3] has analysed a two-element redundant system with generalized repair times and obtained the mean time to system failure. Kumagai[4] has considered the case of a system having two repair facilities and being subject to general failure and exponential repair time distributions. Birolini[5] has analyzed a two-unit redundant system using regenerative processes. Sharma et al.[6] have discussed the transient behaviour of multiple unit systems. All these studies have been made under the assumption that either the failure or the repair time follows exponential distribution. However, in most of the practical situations, both repair and failure distributions follow an arbitrary distribution. Srinivasan and Gopalan[7] have analysed a two-unit cold-standby system with general failure and repair time distributions using a supplementary variable technique in connection with the reliability and the availability of the system. However, the general analysis of an n-unit cold-standby system with general failure and repair time distributions remains open. The main aim of this article is to analyse an n-unit cold-standby system with general failure and repair time distributions. The following assumptions are made for analysing the system: q Initially one unit is switched on and the remaining (n-1) units are in standby. q When the operating unit fails, one unit from the standby is switched on and the unit that has failed is taken up for repair. q Switching is perfect and instantaneous. q The failure and the repair times are random and are arbitrarily distributed.