With the development of statistical and computational technology, landmark-based morphometric analysis has widely been used for species classification and identification in recent years, but it has never been used to identify wild from geographically different basins. Therefore, this research was resized to identify different geographical populations of wild collected from the Tumen River (TR), Liao River (LR), Huang River (HR), Yangtze River (YR), Ou River (OR), and Min River (MR). Landmark-based geometric morphometrics analysis was used to compare the morphological characteristics of their carapaces. The main step was to extract coordinate values of landmark-based points on the carapace, and then perform the relative warp principal component analysis and discriminant analysis to distinguish the 6 wild populations, and finally visualize the morphological variation with thin plate spline analysis and grad distortion. The results showed that:(1) in principal component analysis, wild populations could be divided into 6 relatively independent groups by the first two principal components, though the plots slightly overlapped between different populations; (2) the grad distortion map showed that morphological differences of various populations were mainly manifested in the frontal and lateral spines of the carapace; (3) the discriminative accuracy of various populations in the stepwise discriminant analysis was 80%-96% with a certain risk of misjudgment; (4) cluster analysis showed that YR, HR, OR, and MR individuals clustered independently, whereas TR and LR individuals formed another distinct cluster, whether females or males. In summary, landmark-based morphometrics analysis is an effective means to distinguish wild from different basins, but it still needs to be confirmed by traditional morphometrics analysis and genetic analysis.