This paper presents an efficient and reliable method for general semiconductor device parameter extraction. The parameter extraction technique consists of minimizing a nonlinear objective function. A hybrid approach based on strategical combinations of the Gauss method and a novel minimum search scheme is employed. Various of strategies are proposed to make the parameter extraction task more efficient and accurate. According to the proposed technique, a model-independent parameter extraction program was implemented and highly satisfactory results have been obtained. Experimentally, it is found that, if the initial guess deviates from the optimum solution within 70 percent, a unique set of optimum device parameters can be extracted in five to eight iterations. The parameter extraction technique presented in this paper can be readily applied to sensitivity analysis of device and circuit performances and to quantitative evaluation of model accuracy. It is shown that the presented method is a very useful tool for device and integrated circuit design.