This paper presents a method to be used on correcting finite element models by refering their modal parameters to experimental modal data. The proposed method is based on both the sensitivity analysis and the least square method, and gives the physically realistic solution of design variables, in contrast with the methods based on pseudo inverse. Analytical simulations of a beam and a plate are presented to demonstrate the effectiveness of the proposed method in comparison to the pseudo inverse method in the aspect of structural reasonableness. In addition, the correction of a finite element model of an actual structure (bar) is shown by using experimentally obtained modal parameters.