Agriculture sector has been affected by severe droughts during recent years in Iran. The existence of an agriculture drought warning system can be a useful means for preparedness and to reduce the consequent losses. This study develops a drought agricultural drought vulnerability assessment model using statistical and intelligent methods for Kermanshah province. The model is specific to rainfed wheat and barley and is able to assess real-time agricultural drought vulnerability associated with crop yield losses at critical phenological stages during the growing season. The model inputs are a combination of drought indices including: PDSI, Z-index, CMI, SPI and EDI, using genetic algorithm method. The model results show that the accurate rates are lower at the first stage than those during the later stages. Enhancement of the model with a Geographic Information System (GIS) has also made the model more capable for spatial analysis.