The application of optimization to the solution of practical structural engineering problems has been very limited, despite the great development of the techniques. One of the main reasons for this is the complexity of the generated models, which employ nonlinear functions and generate a space of nonconvex solutions. In addition, most design variables, as in the case of steel profiles, can only assume discrete values. Traditional methods of mathematical programming are very limited in solving problems with these characteristics, opening space to the usage of heuristic methods and, more specifically, to metaheuristic methods. The main advantage of this class of methods is the fact that they only involve values of the functions in the optimization process, regardless of the existence of unimodality or even continuity in the derivatives of the functions involved. The present work presents the application of a heuristic method, the simulated annealing method, to the optimization of steel trusses with parallel chords, also called flat trusses. Initially, several usual configurations of trusses were analyzed, aiming to identify which led to the lowest cost. The top chord coordinates were also included as design variables, with the purpose of verifying the relations between span and height normally indicated in the literature as the most economical. Among the results, it was verified that the inclusion of the height of the truss among the design variables can lead to a significant additional cost savings.