Previous system reliability apportionment models assume that component or subsystem reliability values are exact. However, they are usually estimated from life tests, and therefore, uncertainties are inevitably involved in them. This thesis presents a new system reliability apportionment model considering those uncertainties for the situation where the designer must choose an alternative among several ones in each subsystem. The optimization problem is formulated as a multi-objective mathematical program in which the estimated system reliability is maximized and its variance is minimized at the same time under the resource constraint. Two heuristic algorithms are developed for the above mathematical program and example problems are also solved."