This paper establishes a multi-objective reactive power optimization model based on fuzzy membership functions. A new adaptiveε -MOEA approach is employed to obtainε -approximate Pareto set. The optimal resolution is obtained by a post-evaluation fuzzy logic controller. Compared to other post-evaluation methods， this approach features fast calculation speed， good orientability and self-adaption. Tests on IEEE 14， IEEE 30 and IEEE 118 system demonstrate that the method achieves high efficiency while retains a good quality in solution compared to other methods.