The skulls of 33 extant cat species were characterized through three-dimensional geometric morphometrics using 20 landmarks. A principal component analysis (PCA) was performed with Procrustes fitted coordinates, and the PC-scores were phylogenetically corrected by independent contrasts method. Three PCs allowed for the definition of five cat skull patterns. PC1: ‘snouted/massive-headed cats’ (genus Panthera) opposing the ‘round-headed small cats’ (genus Oncifelis, Prionailurus rubiginosus, Prionailurus bengalensis, among other small cats); PC2: ‘tapering-headed cats’ (Neofelis nebulosa, Herpailurus yagouaroundi, Prionailurus planiceps) opposing the ‘stout-headed cats’ (Acinonyx jubatus, Uncia uncia, Otocolobus manul, Felis margarita, and Felis nigripes); and PC3: ‘low profiled-headed cats’ (mostly, Pr. planiceps). A sixth pattern, the ‘generalized skull’, observed in the Caracal lineage, genus Lynx, Leopardus pardalis, and Catopuma temminckii, indicates a morphological convergence among midsized-cats. The morphological trends ‘snouted/massive’ and ‘round’ clearly denote a co-evolution between size and shape. The other skull patterns evolved unrelatedly to the size (i.e. their allometric variations are not a size function). Nevertheless, each species comprises an amalgam of these patterns, so the influence of the size permeates, in some extent, the skull morphology along all cat lineages. The felid ecomorphological fit to hypercarnivory is obvious; however, different skull shapes in same-sized species with similar habits, indicate that the variation in the skull morphology may result from phenotypic fluctuations, whose adaptive value (if indeed there is any) is still obscure. © 2011 The Linnean Society of London, Biological Journal of the Linnean Society, 2011, 103, 176–190.