An optimum design problem is treated to determine a frame structure minimizing the structural cost or weight under the constraints on the allowable failure probabilities of critical sections specified in the structure. The configuration of the structure and loading conditions together with their probabilistic nature are assumed to be given. The design variables are the geometrical dimensions of the elements. The failure criterion of the critical section is expressed as a function of the strength of the element and the internal force, which is generated by a matrix method. The optimum design problem is reduced to a nonlinear programming problem. The solution to the optimization problem is obtained by applying the sequential linear programming, and numerical examples are given to illustrate the applicability of the proposed method.