Morphology of two Leiognathus brevirostris stocks was analyzed by multivariate and geometric morphometrics. The first two principal components (76.29%) could obviously identify two stocks from Yangjiang and Zhanjiang in principal components analysis by multivariate morphometrics. The first three principal components (64.03%) could also obviously identify two stocks in relative warps anlysis. Morphology difference was mainly on the head of Leiognathus brevirostris from grid deformation. Compared to traditional multivariate morphometrics, landmark method avoided uneven and redundant aspects of form, and allowed graphic reconstruction of shape to investigate minute shape variation among objects of study.