Premise of research: The morphology of leaves is shaped by both historical and current selection acting on constrained developmental systems. For this reason, the phylogenetic signal of these characters is usually overlooked. Methodology: We investigate morphology of the leaflets of all genera of the Zamiaceae using multiple microscopical techniques to test whether leaf characters present a phylogenetic signal and whether they are useful to define clades at a suprageneric level. Pivotal results: Our investigation shows that most genera are quite uniform in their leaflet anatomy, with the largest genera (Zamia, Encephalartos) presenting the highest degree of variation. Using both Bayesian and parsimony methods on two different molecular scaffolds, we are able to show that leaflet anatomy has a strong phylogenetic signal in the Zamiaceae and that many clades retrieved by molecular analyses present potential synapomorphies in their leaflet anatomy. Particularly, the placement of Stangeria in a clade with Zamia and Microcycas is supported by the presence of both an adaxial and an abaxial girder sclerenchyma and the absence of sclerified hypodermis. The placement of Stangeria as sister to Bowenia, on the other hand, is not supported by our analysis. Instead, our results put into question the homology of the similar guard cell morphology in the two genera. Conclusions: We show that leaflet anatomy has a substantial amount of phylogenetic signal in the Zamiaceae, supporting relationships that are not supported by general morphology. Therefore, anatomical investigation represents a promising avenue for plant systematists.