In this paper a structural optimization technique based on a genetic algorithm (GA) is presented. The technique is developed to deal with discrete structural optimization of 2D steel frame structures. Also from a structural point of view, the paper explores the maximum ratio of the effective buckling length when using the finite element approach and that by the British Standard (BS 5950) approach. In order to consider realistic steelwork design problems, the developed technique has been linked to a system of structural design rules (British Standards BS 5950 and BS 6399), interacting with a finite element package (ANSYS). In the formulation of the optimization problem, the objective function is the maximum ratio of the effective buckling length of a column evaluated by FEM and that by BS5950. The cross-sectional properties of the structural members, which form the set of design variables, are chosen from two separate catalogues (universal beams and columns covered by British Standards BS 4). The constraints are imposed on the design criteria stipulated by BS 5950. Two 2D steel frame structures having different number of design variables and subjected to multiple loading cases are analyzed. These examples show that the developed technique based on GA can be successfully incorporated in discrete structural optimization problems of steel frame structures.