A　new　method　is　proposed　to　improve　the　problem　that　many　samples　and　long　search　time　are　required　in　the　traditional　methods　to　determine　the　threshold　factor　of　constant　false　alarm　ratio　(CFAR)　detector.　The　method　estimates　threshold　factors　for　radar　CFAR　detectors　based　on　particle　swarm　optimization　(PSO)　algorithm.　The　problem　to　determine　the　threshold　factor　under　given　false　alarm　probability　is　first　converted　into　a　minimization　problem,　then　the　minimization　problem　is　solved　using　PSO　method.　The　inertia　weight　is　adaptively　adjusted　with　the　iteration　during　minimization.　Therefore　the　method　has　better　performance　of　global　search　and　local　search,　and　higher　efficiency　of　search.　Simulation　results　show　that　the　method　can　accurately　estimate　the　threshold　factor　for　single　radar　or　radar　netting　system.　Compared　with　a　genetic　algorithm,　50　percent　of　running　time　is　saved,　and　the　accuracy　of　estimation　is　improved.　The　numerical　results　indicate　that　the　proposed　scheme　can　quickly　and　accurately　find　the　estimation　of　threshold　factors　for　most　CFAR　detectors.