We conducted a geometric morphometric analysis to investigate the morphological variation of the golden wolf, Canis lupaster, and to clarify the morphological and taxonomic affinities of different taxa of the genera Canis and Lupulella. We suggest that the variation observed within the complex of Canis lupaster may be incompatible with what would be expected for a single species. We hypothesize that the nominal form C. l. soudanicus is a synonym of Lupulella adusta rather than being part of the golden wolf complex. The subspecies C. l. bea has a generalized jackal morphology (i.e., clusters together with L. mesomelas and C. aureus) and C. l. lupaster occupies an intermediate morphospace position, between jackal-like forms and wolf-like forms. These results contrast with previously published molecular analysis in which mitochondrial data failed to identify differences among golden wolf populations, and nuclear evidence points to the existence of groups that are incompatible with those recovered by morphological analysis. Regarding other jackals, our results depict the absence of morphological overlap between L. m. mesomelas and L. m. schmidti and no differences between putative subspecies of L. adusta. We call attention to the need for more integrative approaches to solve the taxonomic questions in various African Canidae.