One of the most important practical considerations in the optimization of steel structures is that structural members are generally to be selected from available steel pro®les. Furthermore, the solutions produced according to the national speci®cations are undoubtedly valuable from the point of view of applicability in real-life practice. This paper reports the development of a computer-based systematic approach for discrete optimal design of planar and space structures composed of one-dimensional elements. The main characteristic of the solution methodology is the use of a genetic algorithm (GA) as the optimizer. Applications and experience on steel frame and truss structures are discussed. The results of comparative studies of the GA against other various discrete and continuous optimization algorithms for a class of representative structural design problems are reported to show the eciency of the former. It is observed that a GA often ®nds the region of the search space containing the global optimum, but not the true optimum itself. Also, in this study an approach based on a proposed multilevel optimization is tested and proved to overcome this shortcoming.