Grain　logistics　is　an　important　logistical　support　of　national　economic　development.　For　a　long　time,　grain　logistics　in　China　more　use　the　way　of　package　transport　which　is　inefficient　and　wasteful.　In　recent　years,　China　gradually　began　to　carry　out　the　transport　way　in　four　bulks　mode　around,　and　achieved　some　success.　This　paper　combines　the　method　of　quantitative　and　qualitative　and　predict　the　amount　of　bulk　grain　logistics　through　waterway　in　Fujian　next　two　years.　Based　on　the　multiple　indicator　selection,　the　paper　establishes　BP　neural　network　prediction　model　and　use　MATLAB　for　quantify　achieve.　Comparison　with　the　results,　the　paper　could　provide　some　useful　advises　for　future　four　bulks　mode　engineering.