This paper presents an Optimized Rational Fraction Polynomial (ORFP) approach for modal parameters estimation from the measurements of the Frequency Response Function (FRF). Although this approach is based on the Rational Fraction Polynomial (RFP) technique described in [1], it suggests the use of a constrained optimization scheme rather than the Forsythe method to overcome the shortcomings of the Forsythe method. The latter are the estimation of modal parameters that do not necessarily describe a stable system and the estimation of fictitious natural frequencies. The formulation of the constrained optimization problem is presented and discussed. The assessment of the performance of the ORFP approach showed that it is better than the RFP approach in terms of its ability to identify modal parameters that ensure a stable system and its flexibility in selecting and setting the natural frequencies of the system. Several illustrative examples are given to demonstrate the robustness of the ORFP approach.