Abstract: An improved adaptive simulated annealing(SA) algorithm, which combined with the feasibility based rule and direct search simulated annealing, was proposed to optimize the frequency characteristics of composite laminates. The optimization goals of laminates are fundamental frequency, frequency band gap and laminate thickness of band gap constraint. The discrete design variables include fiber orientation angles and the layer number. A new point generation mechanism with the dynamic variation search radius improves the drawback of the direction simulated annealing (DSA) algorithm easy to fall into local minima. Introducing the feasibility based rule into SA enhances the efficiency and simplicity of solving constrained problem. Accounting for bending-twisting coupling effects, the Ritz method was employed to calculate the laminate frequency response. Three types of examples of design symmetrically composite laminates with various layer numbers, angle increment and aspect ratios were conducted. Numerical results indicate that the present improved simulated annealing algorithm can effectively solve laminates frequency optimization and provide with more or better global optimum solutions for the stacking sequence.