When　forecasting　the　traffic　demands　of　multiple　modes　of　transportation,　including　urban　rail　transit,　inconsistencies　often　exist　between　the　model　parameters　of　model　split　and　traffic　assignment.　A　classification　method　of　traffic　modes,　which　considered　the　interrelationships　among　various　traffic　models　as　well　as　their　interactions,　is　presented.　Based　on　this　classification　method,　an　expanded　combined　model　split　and　traffic　assignment　model　of　multinomial　modes　is　established,　which　contains　the　disaggregate　model.　The　corresponding　identification　and　algorithm　are　also　provided.　An　example　calculation　is　employed　to　verify　the　model.　The　results　indicate　that　the　model　can　not　only　satisfactorily　forecast　traffic　demand　including　three　kinds　of　traffic　modes,　but　also　can　solve　the　inconsistency　problem　of　the　parameters　at　the　stages　of　model　split　and　traffic　assignment.