This　research　sought　to　establish　an　eco-driving　training　system　based　on　a　driving　simulator.　The　eco-driving　training　system　contained　five　modules:　human　machine　interface,　data　management,　scene　management,　mode　management,　and　evaluation　algorithm　management.　It　was　proposed　to　base　the　new　eco-driving　training　system　on　drivers＇　individual　characteristics.　This　system　first　asked　drivers　to　conduct　a　diagnostic　drive　on　a　stretch　of　roadway　in　a　driving　simulator.　The　data　on　each　driver＇s　non-ecological　driving　behaviors　under　different　events　were　collected.　Then　each　driver　was　given　a　customized　training　course　based　on　an　evaluation　of　his/her　driving　behaviors　during　the　diagnostic　drive.　This　training　process　is　called　eco-driving　training　based　on　individual　characteristics　(EDTIC).　Eighty　taxi　drivers　were　recruited　and　divided　into　two　groups　for　eco-driving　training.　One　group　was　trained　by　watching　videos,　and　the　other　was　trained　by　the　EDTIC　training.　An　analysis　of　results　shows　that　the　EDTIC　training　was　significantly　more　effective　than　traditional　video　training.　Under　the　EDTIC　training,　all　driving　behaviors　improved　and　emissions　and　fuel　consumption　were　greatly　reduced;　the　reduction　was　as　great　as　8.3-8.4%.　The　EDTIC　training　was　proven　effective　at　improving　the　eco-driving　behavior　of　taxi　drivers　(i.e.,　professional　drivers),　and　it　could　also　be　employed　to　train　other　professional　drivers　(bus　and　truck　drivers)　and　non-professional　drivers.