This paper uses an ant system (AS) meta-heuristic optimization method to solve the problem of structure optimization of series-parallel production systems. In the considered problem, redundant machines (elements) and buffers in process are included in order to attain a desirable level of reliability. A procedure which determines the minimal cost system configuration is proposed. In this procedure, multiple choices of producing machines and buffers are allowed from a list of product available in the market. The elements of the system are characterized by their cost, estimated average up and down times, productivity rates and buffers capacities. The reliability is defined as the ability to satisfy the consumer demand which is represented as a piecewise cumulative load curve. The proposed meta-heuristic is used as an optimization technique to seek for the optimal design configuration. The advantage of the proposed AS approach is that allows machines and buffers with different parameters to be allocated.