Glutamine phosphoribosylpyrophosphate amidotransferase is stable in growing cells, but is inactivated in an oxygen-dependent process at various rates in starving or antibiotic-treated cells. On the basis of studies of the purified enzyme, we suggested (D.A. Bernlohr and R.L. Switzer, Biochemistry 20:5675-5681, 1981) that the inactivation in vivo was regulated by substrate stabilization and a competition between stabilizing (AMP) and destabilizing (GMP, GDP, and ADP) nucleotides. This proposal was tested by measuring the intracellular levels of these metabolites under cultural conditions in which the stability of the amidotransferase varied. The results established that the stability of amidotransferase in vivo cannot be explained by the simple interactions observed in vitro. Metabolite levels associated with stability of the enzyme in growing cells did not confer stability under other conditions, such as ammonia starvation or refeeding of glucose-starved cells. The data suggest that a previously unrecognized event, possibly a covalent modification of amidotransferase, is required to mark the enzyme for oxygen-dependent inactivation.