The 2 cryptic species of multimammate mice that occur widely in South Africa were, for many decades, lumped in a single species Mastomys natalensis sensu lato. This taxon was intensively studied because these rodents play a role in the epidemiology of various zoonoses (notably bubonic plague), serve as important models for biomedical research, and also sometimes cause extensive agricultural damage. The discovery that M. natalensis sensu lato includes 2 cryptic species in southern Africa has complicated interpretation of the results of past research because previous studies may have been based on either M. coucha or M. natalensis, or even on specimens of both species. A previous study aimed at craniometrically distinguishing between these species in South Africa met with some success, but excluded subadult specimens (which constitute a large proportion of museum collections), and cautioned that intraspecific variation might hamper interspecific discrimination. We therefore studied the nature and extent of nongeographic craniological variation using 2 populations of each species. The discriminatory value of a putative diagnostic dental character (presence or absence of t3 cusp on M3) is limited because of considerable variation within and among populations of M. coucha. Sexual dimorphism is negligible in both species, whereas age-related variation is pronounced and involves mainly differences in overall size, as well as some subtle differences in the configuration of the interorbital, postpalatal, and dental regions. Multiple discriminant function analyses aimed at craniometrically distinguishing between the 2 species showed that the inclusion of subadult specimens reduced a posteriori classification accuracy below 95% confidence levels, implying that age-related variation is sufficiently pronounced to obscure interspecific craniometric differences. Future studies aimed at craniometrically distinguishing between these cryptic species will, therefore, have to statistically correct for age-related variation.