Based on the genetic clustering from 42 microsatellite (SSR) markers with a combination of their geographic origin and germplasm characteristics, 124 maize landraces from Wuling Mountain region in China were used for constructing a core collection. Four evaluating parameters for maize landrace core collection, including mean difference percentage (MD), variance difference percentage (VD), coincidence rate of range (CR), and variable rate of coefficient of variation (VR), were assessed with 20 quantitative traits. It was shown that genetic relationships among landraces in Wuling Mountain region had the tendency to associate with their geographic origins. The 124 landraces were clustered into 18 subgroups when the coefficient of genetic similarity (GS) is 0.28. Eighteen landraces, each of which was from one subgroup, were applied to construct the core collection with a sampling percentage of 15%. Comparison of the initial and core collection indicated that there existed no significant differences in most quantitative traits. An average of 6.3 and 6.5 alleles were detected in the initial and core collection, respectively. Mean polymorphism information content in the core collection (0.75) was higher than that in the initial one (0.72). MD was lesser than 20% and CR was more than 80%. The results showed that the sampling strategy would be feasible for constructing the core collection that well represents the genetic diversity of the initial one. Based on the genetic clustering from 42 microsatellite (SSR) markers with a combination of their geographic origin and germplasm characteristics, 124 maize landraces from Wuling Mountain region in China were used for constructing a core collection. Four evaluating parameters for maize landrace core collection, including mean difference percentage (MD), variance difference percentage (VD), coincidence rate of range (CR), and variable rate of coefficient of variation (VR), were assessed with 20 quantitative traits. It was shown that genetic relationship among landraces in Wuling Mountain region had the tendency to associate with their geographic origins. The 124 landraces were clustered into 18 subgroups when the coefficient of genetic similarity (GS) is 0.28. Eighteen landraces, each of which was from one subgroup, were applied to construct the core collection with a sampling percentage of 15%. Comparison of the initial and core collection indicates that there existed no significant di Mean polymorphism information content in the core collection (0.75) was higher than that in the initial one (0.72). MD was lesser than 20% and CR was more than 80%. The results showed that the sampling strategy would be feasible for constructing the core collection that well represents the genetic diversity of the initial one.