Abstract: 【Objective】 Screening of High Polymorphism SSR Primers for Classification of Heterosis Group in Maize (Zea mays L.). 【Method】 A total of 160 pairs of SSR primers, including the core primers used in the past, which evenly distributed on 10 linkaged groups in maize, were selected, and then were amplified in DNA of 104 inbred lines.And then, based on the PIC values of the primers and their distribution on chromosome, 40 pairs, 30 pairs, 20 pairs and 10 pairs of high polymorphic primers were taken to establish 4 sets of new primer systems to classify the 104 inbred lines. 【Result】 (1) Of the 160 pairs of primers, 63 were highly polymorphic, in which the highest PIC value was 0.762 that were close to or higher than that of core primers used in the past.The core SSR commonly used in the past retained only about 40% of the 63 primers.(2) In the 4 sets of primer systems, the classification of 40 pairs and 30 pairs of primer system were highly consistent with the groups of inbred lines.20 pairs of system were 90.5% consistent with that of 40 pairs of primers, and the system of 10 pairs of primers were only 81.0% consistent.(3) When 40 pairs of primers were used to classify 104 inbred lines, five groups were distinguished, that is, Reid, improved Reid, Lancaster, improved Huang and Lvda-Red-Cob, respectively, which was consistent with the heterotic groups used in our country at present.In each group, the most big distantly sub-group from main group consisted of the inbred lines derived from the hybridization between different groups.(4) Population genetic structure component of representative inbred lines of each group was single structure component, and the lines of subgroup were multi structure component. 【Conclusion】 The decrease of polymorphism of some SSR primers was observed in the new inbred lines. The new system established by 40 pairs of primers has accurate classification function and could be directly applied to group classification.For the system of 20 pairs of primers, its accuracy is slightly lower, but its workload is halved, which is useful in the preliminary classification of batch inbred lines.