We apply a selection and estimation procedure for the detection of multiple targets in clutter. The selection and estimation (SE) test perfoms better than a fixed order statistic (OS) detector. The SE test has some similarity to a variable trimmed mean (VTM) test and has a slight performance advantage. Unlike the VTM, whose performance is evaluated only by simulation studies, the performance of the SE test, for Rayleigh fluctuating target model, can be evaluated analytically. Further improvement in the performance is possible if a better selection procedure can be found.