It　is　hard　to　describe　driver＇s　behavior　with　precise　algorithm　because　of　its　fuzzy　and　indeterminable　characteristic　when　the　driver　follows　a　leading　car.　However,　it　is　suitable　to　apply　fuzzy　neural　network　(FNN)　to　study　driving　behavior　due　to　its　good　characteristic　of　self-study　and　easily　expressing　common　control　experience.　The　paper　uses　RD/ED　and　RS　as　two　inputs　and　the　following　car＇s　acceleration　as　the　output　and　simulates　in　MATLAB,　model　validation　results　have　shown　that　the　FNN　model　can　replicate　the　driving　behavior　of　real　systems　in　road　cases.