A number of observations in the Escherichia coli and Salmonella typhimurium literature could be explained by the hypothesis that a particular purine ribonucleotide precursor can be converted to the corresponding deoxyribonucleotide triphosphate, thereby becoming a base-analogue mutagen. The metabolite in question, AICAR (5-amino-4-carboxamide imidazole riboside 5'-phosphate), is also a by-product of histidine biosynthesis, and its (ribo)triphosphate derivative, ZTP, has been detected in E. coli. We constructed E. coli tester strains that had either a normal AICAR pool (pur+ his+ strains cultivated without purines or histidine) or no AICAR pool (purF hisG mutant strains, lacking the first enzyme of each pathway and cultivated in the presence of adenine and histidine). Using a set of lacZ mutations, each of which can revert to Lac+ only by a specific substitution mutation, we found that no base substitution event occurs at a higher frequency in the presence of an AICAR pool. We conclude that the normal AICAR pool in E. coli is not a significant source of spontaneous base substitution mutagenesis.