Using phenotypes to explore and describe biological diversity has become less popular than using genetics to do so. Results from the two approaches often conflict at the species level and below, the very ground floor of biodiversity. However, because in today's data sets phenotypic divergence is probably driven mostly by selection and genetic divergence by stochastic processes, we should not expect them to be tightly coupled at population-to-species evolutionary depths. For heuristic purposes it is useful to consider phenotypic and genetic data as largely unidimensional axes in an inherently multidimensional process, and this is perhaps the source of the controversies surrounding each approach. Phenotypic andgenotypic data sets might give very different portrayals of evolutionary trajectories in adaptive and nonadaptive space. Integrating these data sets provides a roadmap for theoretical and empirical research. The exploration of the multidimensional relationships of the two types of differentiation in diverging populations is providing important insights both into the units of biodiversity and into the processes responsible for their generation.