Leaves can be a useful source of taxonomic information in plants particularly when flowers and fruits are absent during certain periods of the year. In this study, we applied an elliptic Fourier analysis (EFA)-based morphometric technique to assess leaf morphological divergence among four species of southern African Strychnos section Densiflorae. Using leaf specimen images from field and herbarium collections, we extracted six shape variables [i.e. principal components (PCs)] from the Fourier coefficients and used these variables to describe leaf outline among the species. Our results indicate that the symmetric component of a leaf is the main source of shape differences accounting for 90.25% of total leaf shape variation and captures the more obvious range of observed shapes. PC1 of the symmetric variables describes a wide range of visually observable leaf shape among the species. MANOVA revealed significant interspecific differences except between S. innocua and S. madagascariensis, which could not be separated by outline analysis. A cross-validated group classification suggests that S. gerrardii, with a classification rate of 88.4%, is distinct from S. madagascariensis, contrary to some taxonomic treatments. We discuss the value of geometric morphometrics at detecting subtle morphological variations and the evolutionary implications of such variations, which may be undetectable to the human eye. © 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2012, 170, 542–553.