To solve the low accuracy problem of the template matching(TM) method under condition of feature incomplete database in traditional EW, a multiple division support vector machine (MD-SVM) jamming decision method is proposed. For the air-to-air scene airborne multi-functional fire control radar, a feature incomplete database is constructed. The sample set is divided into multiple sample subsets using multiple division method. The inner product of multiple sample subset spaces is transformed to the inner product of the feature space by SVM. Feature space hyperplane is established, and the jamming tags corresponding to the sample subsets are output. Then the jamming style is quickly and effectively determined. The experimental results show that the proposed method can effectively improve the accuracy and robustness of the jamming decision compared with the traditional TM and non-multiple division method. The feature incomplete database jamming decision problem is solved for its excellent learning and generalization ability.