Abstract-This paper proposes a new approach for efficiently determining the unwanted interfering samples in the reference window, for the ordered statistics constant false alarm rate detector, based on the application of the information theoretic criteria principle. The proposed processor termed as Forward Automatic Order Selection Ordered Statistics Detector (FAOSOSD) does not require any prior information about the number of interfering targets. The proposed design aims to improve the Ordered Statistics Constant False Alarm Rate detector performance under severe interference situations. The number of interfering targets is obtained by minimizing the information theoretic criteria. Simulation results that illustrate the performance of the proposed method versus the classical OS-CFAR, the AND-CFAR and the OR-CFAR detectors are presented and discussed.