One of the enduring temptations of evolutionary theory is the extrapolation from short-term to long-term, from a few species to all species. Unfortunately, the study of experimental evolution reveals that extrapolation from local to general patterns of evolution is not usually successful. The present article supports this conclusion using evidence from the experimental evolution of life-history in Drosophila. The following factors demonstrably undermine evolutionary correlations between functional characters: inbreeding, genotype-by-environment interaction, novel foci of selection, long-term selection, and alternative genetic backgrounds. The virtual certainty that at least one of these factors will arise during evolution shreds the prospects for global theories of the effects of adaptation. The effects of evolution apparently don't generalize, even though evolution is a global process.