Abstract: A gradient based optimization method for variable stiffness design of laminates was proposed. The optimal element density and orientations were determined in the optimization process. In order to attain industrial relevance, the manufacturing constraint should be satisfied and candidate orientations were limited to a finite set. For the purpose of eliminating the number of constraints and design variables in optimization model, the density distribution curve method (DDCM) was established to parameterize element density. The elastic matrix of element was determined on basis of interpolation scheme in bi-value coding parameterization(BCP) method, according to element density and design variables of orientations. Strucrure compliance and volume were treated as optimization objective and constraint respectively in optimization process. The optimization problem was solved by the convex programming dual algorithm. Numerical examples for the carbon fiber composite demonstrate that reasonable results can be got by DDCM with relatively high convergence rate.