A　simulation　of　bi-direction　pedestrian　flow　based　on　cellular　automata　(CA)　will　be　presented　from　two　aspects:　direction　split　and　pedestrians＇　walking　habit　in　this　paper.　The　Simulation　uses　Dynamic　Parameters　Model　(DPM)　to　simplify　tactically　the　decision-making　process　of　pedestrians　in　their　movements.　A　new　parameter　right-hand　parameter　is　introduced　to　describe　the　pedestrians＇　walking　preference.　The　relationships　of　velocity-density　and　flow-density　will　be　studied　and　analyzed.　It　is　found　that　there　are　phase　transitions　at　the　critical　density　point,　and　the　pedestrian　flow　shows　distinctive　characteristics　at　different　phases　with　different　relationships　of　velocity-density　and　flow-density.　It　is　also　found　that　direction　split　and　pedestrians＇　walking　habit　affect　the　value　of　critical　density　point　and　the　figures　of　velocity-density　and　volume-density　curves.　In　conclusion,　the　simulation　can　reflect　and　describe　some　pedestrian　flow　self-organization　phenomena　and　transition　trend　of　empirical　pedestrian　flow　curves.　(C)　2009　Elsevier　B.V.　All　rights　reserved.