The invention discloses a method for predicting electric power material agreement inventory based on big data. The method comprises the steps: collecting detail plans and delivery data of various materials; collecting supply periods, supply timeliness ratios and accuracy data of various materials; establishing a general material supply and reserve data model according to the requirements and supply distribution of various materials; and predicting the material inventory according to the material supply and reserve data model. According to the invention, the supply of materials is effectively ensured, and the problems of material shortage and resource waste caused by too many purchased materials are prevented, and the maximum utilization of electric power materials is realized.