The generation of mutants in model organisms by geneticists and developmental biologists over the last century has occasionally produced phenotypes that are startlingly reminiscent of those seen in other species. Such extreme mutations have generally been dismissed by evolutionary geneticists since the "modern synthesis" as irrelevant to adaptation and speciation. But only in recent years has information on the molecular bases of mutant phenotypes become widely available, and thus work on testing the relevance of such extreme mutations to the generation of phylogenetic diversity has just begun. Here we evaluate whether evolutionary mimics are, in fact, useful for pinpointing the genetic differences that distinguish morphological variants generated during evolution. Examples come from both plants and animals, and range from intraspecific to interordinal taxonomic ranges. The use of mutationally defined candidate genes to predict evolutionary mechanisms has so far been most fruitful in explaining intraspecific variants, where it has been effective in both plants and animals. In several cases these efforts were facilitated or supported by parallel results from quantitative trait loci studies, in which natural alleles controlling continuous variation in developmental model organisms were mapped to mutationally defined genes. However, despite these successes the approach's utility seems to rapidly decay as a function of phylogenetic distance. This suggests that the divergence of developmental genetic systems is great even in closely related organisms and may become intractable at larger distances. We discuss this result in the context of what it teaches us about development, the future prospects of the candidate gene approach, and the historical debate over process in micro- and macroevolution.