Abstract: A structural optimization model is developed,in which the kinematic dexterity of a 6 DOF Stewart paralllel manipulator is considered as the objective function, and the design space,the ratio of the maximum to minimum length of each leg,the limit angles of universal and spherical joints are taken as constraints.The Metropolis genetic algorithm,in which the Metropolis criterion of simulated annealing algorithm are incorporated in the selection operations of real-coded genetic algorithm,is applied to the structural design optimization of parallel manipulator.Results obtained by canonical and Metropolis genetic algorithms are compared,proving the validity and advantage of Metropolis genetic algorithm for structural design optimization of parallel manipulator.