Driver　fatigue　detection　is　attractive　in　Intelligent　Transportation　Systems　(ITS).　In　this　paper,　we　analyze　some　usual　methods　of　driver　fatigue　detection,　and　choose　the　Machine　Vision　Based　method　to　create　a　real-time　driver　fatigue　detection　system.　First　we　separate　the　background　through　skin　color　segmentation,　and　band　the　face　region　in　a　rectangle　according　to　prior　information.　Then　we　locate　eyes　by　using　Haar-Like　features.　Finally　we　propose　a　new　method　to　compute　percentage　of　eyelid　closure　by　using　the　features　of　the　rectangle　border　of　iris.　The　results　of　indoors　experiments　show　good　detection　rate　and　precision.