Nucleic acids are replicated with conspicuous fidelity. Infrequently, however, they undergo changes in sequence, and this process of change (mutation) generates the variability that allows evolution. As the result of studies of bacterial variation, it is now widely believed that mutations arise continuously and without any consideration for their utility. In this paper, we briefly review the source of this idea and then describe some experiments suggesting that cells may have mechanisms for choosing which mutations will occur.