A hierarchy optimization model was proposed for carbon fiber reinforced polymer laminate with dramatic stress variation. An optimal design with minimal mass was obtained by dividing the structure into several sub-laminates and optimizing the location, size, number of plies and stacking sequences of each sub-laminate with the constraints of strength and manufacturability. The location and size of each sub-laminate were optimized by using the reference ply in the first and second level optimization and the number of plies and stacking sequences of each sub-laminate were optimized by introducing the parametric method of cubic spline interpolation in the third level optimization. The reference ply was designed to reduce the number of design variables and the problem of uncertainty of design variables number was solved by the parametric method of cubic spline interpolation, taking layer angle as design variable. Finite element method was used to analyze the structure mechanically whose strength was judged by Tsai-Wu criterion and genetic algorithm was used in the second and third level optimization to the optimization problem. The example calculation shows that the results of ply hierarchical optimization of ply drop are reasonable and the ply drop can considerably reduce the mass of structures compared with results of uniform layer method.