Several recent reports have claimed that adaptive mutants in bacteria
and yeast are induced by selective conditions. The results of these re
ports suggest that mutants can arise nonrandomly with respect to fitne
ss, contrary to what has been widely accepted. In several cases that h
ave received careful experimental reexamination, however, the detectio
n of seemingly nonrandom mutation has been explained as an experimenta
l artifact. In the remaining cases, there is no evidence to suggest th
at cells have the capacity to direct or choose which genetic variants
will arise. Instead, current models propose processes by which genetic
variants persist as mutations only if they enable cell growth and DNA
replication. Most of these models are apparently contradicted by expe
rimental data. One model, the hypermutable state model, has recently r
eceived limited circumstantial support. However, in this model the ori
gin of adaptive mutants is random; the apparent nonrandomness of mutat
ion is merely a consequence of natural selection. The critical distinc
tion between the origin of genetic variation (mutation) and the possib
le consequence of that variation (selection) has been neglected by pro
ponents of directed mutation.