Despite recent experiments showing that BrdUrd-induced mutagenesis can be independent of the level of bromouracil (BrUra) substitution [Kaufman, E.R. & Davidson, R.L. (1978) Proc. Natl. Acad. Sci. USA 75, 4982-4986; Aebersold, P.M. (1976) Mutat. Res. 36, 357-362], BrUra.G base mispairs are a major determinant of mutagenesis. We propose that the experiments cited above are sensitive predominantly to G . C leads to A . T transitions driven by the immeasurably small but highly mutagenic substitution of BrUra for cytosine and not by the gross substitution of BrUra for thymine in DNA. More generally, we show how accumulated evidence suggests that both BrdUrd and 2-aminopurine have two mutagenic effects intracellularly: perturbation of normal deoxyribonucleoside triphosphate pools and analogue mispairs in DNA. We propose a molecular basis for various observations of normal exogenous deoxyribonucleosides as synergists and counteragents to base analogue mutagenesis. A model is proposed to explain the antipolarity of BrdUrd and 2-aminopurine mutagenesis--i.e., why mutants at hot spots for induction by one base analogue are usually hot spots for reversion by the other. It is concluded that the configuration of the neighboring nucleotides surrounding the base analogue mispair, and not the base analogue's preference for inducing A . T leads to G . C or G . C leads to A . T errors, is responsible for the antipolarity of BrdUrd and 2-aminopurine mutagenesis.