Consider a failure model for a stochastic system. A shock is any perturbation to the system which causes a random amount of damage to the system. Any of the shocks can cause the system to fail at shock times. The amount of damage at each shock is a function of the sum of the magnitudes of damage caused from all previous shocks. The times between shocks form a sequence of independent and identically distributed random variables. The system must be replaced upon failure at some cost but it also can be replaced before failure at a lower cost. The long term expected cost per unit time criterion is used. Structural relationships of the optimal replacement policy under the appropriate regularity conditions will be developed. And these relationships will provide theoretical background for the algorithm development.