This paper presents a method for the shape and topology optimization (i. e., layout optimization)of mechanical structures by using a genetic algorithm. It may be difficult or not always possible to obtain the global optima for layout optimization problems by using the traditional gradient-based algorithms. The generic algorithm is known to be an excellent means to obtain the global optima for the problems. The algorithm, however, usually requires a large number of generations and individuals for the final layouts to be obtained certainly. We propose a method in which the behaviors of layouts are pursued for the optimum layout by selecting the layouts of the mechanical structures as individuals instead of the finite elements, and by adopting removal and addition parameters of elements as chromosomes. The efficiency and reliability of the method are demonstrated with various experiments for finite-element models of two-dimensional continuums and for minimum weight problems.