A　quantitative　analysis　is　conducted　on　the　chaos　phenomenon　in　the　car-following　fleet　based　on　the　headway　kept　by　the　two　consecutive　vehicles　and　the　follow-up　vehicle＇s　speed.　By　studying　the　microscopic　characteristics　of　the　vehicles　fleet　in　the　process　of　stimuli　and　reaction,　it　is　proved　that　there　exists　a　chaos　occurrence　in　the　car-following　fleet.　In　order　to　reasonably　model　the　expected　headway　in　traffic　stream,　four　expected　desired　headway　models　are　proposed.　Then,　the　Rossler　chaos　attractors　are　introduced　in　those　four　models　to　choose　the　models　that　can　describe　the　actual　traffic　characteristics.　The　calibration　and　validation　procedures　are　conducted　using　high　accurate　GPS　equipped　vehicles　to　collect　data　from　expressways.　Based　on　the　results　it　can　be　seen　that　the　modified　headway　models　can　best　reflect　the　car-following　characteristics　of　the　real　traffic　flow.