The araB-lacZ fusion system has been a key case in the 'directed mutation' controversy. Fusions did not occur detectably during normal growth but formed readily after prolonged incubation on selective Ara-Lac medium. To distinguish the roles of starvation stress and selective substrates in coding sequence fusions, we applied sib selection and PCR technologies. Sib selection of the prefusion strain, MCS2, starved under aerobic conditions permitted us to isolate active fusion clones which had never been in contact with arabinose or lactose. Hence, a directive role for selective substrates is not essential. Aerobiosis was necessary for fusions to appear in glucose-starved cultures. The difference in fusion formation between normal and starved conditions is best explained by the response of a signal transduction network to physiological stimuli to activate Mu prophage joining of araB and lacZ sequences. PCR analysis revealed that direct plating on selective Ara-Lac agar yielded mostly a single class of 'standard' fusions, while sib selection yielded a broader spectrum of fusion structures. Standard fusions were found to occur within a narrow 9 bp window in lacZ. The high frequency of standard fusions in glucose-starved cultures suggested efficient and/or specific Mu action.