For traditional CFAR detection, the detectors cannot adapt to homogenous and non-homogenous environment simultaneously, so a modified CAFR detector (IEGOS CFAR) is proposed based on greatest selection in this paper. Using the switching method, the local clutter power estimates in leading window and trailing window were obtained by choosing the reference cells adaptively, then the greatest of them was taken as the total clutter power estimate to realize the CFAR detection. Under the assumption of Swerling Ⅱ target and Rayleigh distribution clutter, the CFAR property was proved. Comparisons between CA, GO, SO ,OS and EXS detectors show that IEGOS owns better performance both in homogenous and non-homogenous environment caused by interference and clutter edge. The detector is suitable for application since no sample ordering is needed.