Crossover is one of the three basic operators in ally genetic algorithm (GA). Several crossover techniques have been proposed and their relative merits are currently under investigation. This paper starts with a brief discussion of the working scheme of genetic algorithms (GAs) and crossover techniques commonly used in previous GA applications. Next, these techniques are tested on two truss size optimization problems, and are evaluated with respect to exploration and exploitation aspects of the search process. Finally, the paper proposes two newly developed crossover techniques, through which a better efficiency of GAs can be obtained. Comparative studies are carried out between the proposed and common crossover techniques, and the results are fully discussed.