To investigate the morphometric differentiation of Harpadon nehereus from coastal areas of China, multivariate mathematical-statistical analysis of 29 meristic characters (199 individuals) of eight H. nehereus populations were conducted by traditional morphological and geometric framework methods. Principal component analysis (PCA) was used to construct seven principal components (PC), with a cumulative contribution ratio of 71.481%. The scatter-plot of the principal components showed that the characteristics of the Zhoushan and Chaozhou populations were different from the others. PC1 reflected differences in the anal fin size and the oblique distance of fish body between the ventral and anal fins. PC2 accounted for the distance between the eyes and the mouth crack length. A one-way analysis of variance (ANOVA) indicated no significant differences in the mouth crack length/head length (MCL/HL), the origin of the adipose fin to the origin of the pelvic fin/body length (D₃-D₈/BL), the origin of the anal fin to the origin of the dorsal fin/body length (D₅-D₆/BL), or the terminus of the anal fin to the terminus of the adipose fin/body length (D₇-D₁₀/BL). Phylogenetic analyses showed that the Zhoushan population clustered in the outermost layer of the tree, followed by the Haikou and Qingdao populations. Stepwise regression analysis was used to select 9 character values of the greatest effects to establish the discriminant formula. The discriminating accuracy of the Qingdao population was 100%, and the comprehensive discrimination rate was 66.9%. The results showed there were some morphological differences among different populations of H. nehereus. The body length, anal fin position and head shape were the dominant characteristics to distinguish them. However, these differences had no corresponding relationships with geographical distribution. It was speculated that the frequent communication between different populations made the morphological differences less affected by geographical distance, and environmental factors may be the main reason.