Existing power system parameter measuring methods often aim at only one error factor， which makes it difficult to obtain accurate results when data sampled from the power system is influenced by several errors. In view of this problem， a parameter extremum optimization model is developed， in which such error factors as the decaying DC components， nonsynchronous sampling and harmonics are precisely depicted. A new measuring method based on the hybrid genetic algorithm（HGA） is developed to use this model for finding accurate power system parameters such as the amplitude， phase， frequency and harmonics. As the ordinary genetic algorithm（GA） converges slowly and the classic method is sensitive to initial guess， the HGA combines the GA operator and hybrid quasi-Newton operator together and， with the GA operator performing global searching in the solution space， the hybrid operator carrying out the strongly local searching at the same time to achieve fast calculation of the power parameters. Simulations show that the HGA method can effectively improve the efficiency and accuracy of power parameter measurement， laying a solid foundation for further monitoring and control of power systems.