Abstract A crucial question in wildlife management concerns the definition of ecologically meaningful population units. For the California sea lion (Zalophus californianus), three population units are recognized in Mexico based on mitochondrial \{DNA\} and geographical distances; these units are thought to be genetically isolated. In the present study, we quantitatively compare patterns of shape variation and evaluate different classification approaches to confirm this regional fragmentation among the Mexican colonies and assess sexual dimorphism in skull morphometry. We employed 20 linear measurements of 368 skulls of specimens from the Mexican Pacific (including the Gulf of California), evaluated correlation among the features, and performed multivariate analyses. To provide robustness to the classification by sex and region, we evaluated the use of kernel-based classifiers. According to skull morphometry, the classification of individuals to sex is very reliable, and there is a phenotypic regionalization among colonies that coincides with the proposed regional population structure. We suggest that inside the Gulf, there is gene flow between females but that it is reduced with geographic distance. The phenotypic differences between the Gulf of California and Pacific for both sexes are consistent with the hypothesis of little or no gene flow between these regions.