This　research　developed　an　eco-driving　feedback　system　based　on　a　driving　simulator　to　support　eco-driving　training.　This　support　system　could　provide　both　dynamic　and　static　feedback　to　improve　drivers＇　eco-driving　behavior.　In　the　process　of　driving,　drivers　could　get　voice　prompts　(e.g.,　please　avoid　accelerating　rapidly)　once　non-eco-driving　behavior　appeared,　and　also　could　see　the　real-time　CO2　emissions　curves.　After　driving,　drivers　could　receive　an　eco-driving　evaluation　report　including　their　fuel　consumption　rank,　potential　of　fuel　saving　and　driving　advice　corresponding　to　their　driving　behavior.　In　this　support　system,　five　items　of　non-eco-driving　behavior　(i.e.,　quick　accelerate,　rapid　decelerate,　engine　revolutions　at　a　high　level,　too　fast　or　unstable　speed　on　freeways　and　idling　for　a　longer　time)　were　defined　and　could　be　detected.　To　validate　this　support　system＇s　effectiveness　in　reducing　fuel　consumption　and　emissions,　22　participants　were　recruited　and　three　driving　tests　were　conducted,　first　without　using　the　support　system,　then　static　feedback　and　then　dynamic　feedback　utilized　respectively.　A　reduction　of　5.37%　for　CO2　emissions　and　5.45%　for　fuel　consumption　was　obtained.　The　results　indicated　that　the　developed　eco-driving　support　system　was　an　effective　training　tool　to　improve　drivers＇　eco-driving　behavior　in　reducing　emissions　and　fuel　consumption.　(C)　2015　Elsevier　Ltd.　All　rights　reserved.