A hierarchical birth-death process is often used as a model for the diversification of species and higher taxa. We evaluate the validity of this null hypothesis by introducing a new measure: the distribution of species within higher taxa. The proportion of monotypic taxa and the number of species in the largest taxon generated by 1000 simulations are compared with extant mammalian data. The initial model is amended to include two sampling patterns, polyphyletic origins as well as ecological and genetic constraints on higher taxon origination rates. Simulated results are extremely variable. For genera, they are found to predict the true distribution of species quite well. For families, however, the null hypothesis is rejected in all its forms. Simulated distributions have both too many monotypic taxa and too large a dominant taxon. The latter finding stands in contrast to previous claims. Interestingly, although polyphyly and constraints slightly improve the fit of the simulations to the data, sampling does not. Finally, Smith and Patterson's dismissal of monotypic taxa is reviewed in light of this model. We argue that while some apparently monotypic taxa are sampling artifacts, this observation has no bearing on the true proportion of monotypic taxa.