|In this study we present a two replacement policies, first based on the number of down times, k of the system such that the long run expected reward per unit time is maximized and second is based on number of failures, N of component 1 such that long run expected reward per unit time is maximized. For the first policy system can be replaced when the number of down times for the system reaches k and for the second policy system can be replaced when number of failure of the component 1 reaches to N . We consider a cold standby system with identical components and failure rate of components are relatively large (overloaded system). Triangular fuzzy numbers are used to represent failure and repair rates as they allow expert opinion, linguistic variables, operating conditions, uncertainty and imprecision in information, to be incorporated into system model. It is assumed that the working time and repair time for the component have independent exponential distribution. The expression for long run expected reward per unit time for a renewal cycle is derived and illustrated proposed policies with numerical examples and simulation study. Also compare two policies by using simulation method.