The orthogonal polynomial method is widely used in the identif ication of modal parameters performed on the frequency response function (FRF) curves. The disadvantage of this method is difficult to determine the number of modes an d the order of the polynomial, while it is difficult to intervene with the identification process by the operator, which can result in the omission of some modes or occurrence of false modes. In this work, based on existing algorithm, t he number of modes and poles of the FRF is first determined by use of the stabil ization diagram and mode indicator function (MIF) in a combination way, and then the residue is estimated by using linear least square method. Furthermore, this method is extended to the global FRF curvefitting. A comparison of the modal parameters estimated from a measured FRF by using existing orthogonal polynomia l algorithm or the commercial software ME′scope VES to those estimated by the m ethod suggested in this work shows that the phenonmena of false modes and the om ission of some modes no longer occur to the results given by the method besides that it can get accurate and stable results, and has great antidisturbance abi lity.