This thesis deals with a redundancy optimization problem for a mixed series-parallel system, where some different types of components are used in each stage so as to incorporate a multi-choice constraint. The objective is to maximize the system reliability subject to various constrains concerned with cost, weight, and volume. The problem is often considered for designing a communication system or a missile-guided weapon system. For the problem, a heuristic algorithm is developed to give a near optimal solution efficiently, and a branch and bound procedure is also exploited for the optimal solution. Numerical examples are solved to illustrate the procedure.