Evenness is the component of species diversity that was built to be mathematically independent from species richness. Yet, earlier work questioned the generality of this independence. However, this earlier work involved only a few relative abundance distributions (RADs), very limited gradients of species richness, or evenness indices, such as Pielou's index, that were not defined in relation to a precise, coherent set of axioms. We show that for very different theoretical RADs, a relationship between evenness indices and species richness does exist and is strong-at least for some RADs and/or when species richness is under 20. Furthermore, this relationship is mostly negative, which contradicts some previous studies. We also show that, for the very uneven RADs we tested, evenness indices depend even more on species richness than diversity indices do. Finally, we discuss the philosophy behind the analysis of evenness-richness relationships within and between RADs, as well as the interest of correcting evenness indices for species richness when species richness varies, especially with many values under 20.