In　this　paper,　a　new　statistic　car-following　regulation　was　described　and　car-following　model　was　established　by　means　of　the　time　delay　differential　equation　of　chaos　dynamics.　Using　a　high　precision　GPS　receiver　we　collected　the　time　serial　data　that　represented　drivers＇　behaviors　and　vehicle　status　in　real　traffic　flow.　A　complete　course　of　car-following　was　divided　up　into　three　stages,　acceleration,　steady,　and　deceleration　stage.　Using　the　dynamic　gathered　measures,　we　analyzed　the　average　acceleration　as　a　compartmentalized　index　to　reveal　the　relationship　between　the　distance　headway　and　the　following　vehicle　velocity　in　an　acceleration　stage　and　deceleration　stage,　the　regression　relationship　between　the　response　time　of　the　following　driver　and　the　following　vehicle　velocity　in　the　deceleration　stage　and　the　regression　relationship　among　expected　distance　headway,　headway　and　following　velocity　by　means　of　mathematical　statistics.　We　made　out　the　relationship　between　the　following　vehicle　acceleration　and　the　distance　headway,　and　the　relationship　between　the　following　vehicle　acceleration　and　the　following　vehicle　velocity　in　a　steady　stage.　And　we　validated　these　relationships　using　simulation　tests.　©　2011　ASCE.