O6-Methylguanine (m6G) was incorporated site-specifically into two 25-base oligonucleotides differing only in the nucleotide on the 3' side of the modified base. Templates were primed with oligonucleotides terminating one or two bases prior to the site at which incorporation kinetics were to be investigated. Escherichia coli DNA polymerase I (Klenow fragment) was used to determine the apparent Km and relative Vmax of incorporation of either dCTP or dTTP opposite m6G or G. These data were used to calculate the relative frequency of incorporation opposite the m6G or the unmodified G. When the sequence was 3'-Cm6G-5', there was a 6- to 7-fold preference for formation of a m6G.T pair compared with m6G.C. The m6G.T frequency, based on Vmax/Km, was at least 50-fold greater than that of a G.T pair at the same site. Changing the sequence to 3'-Tm6G-5' had a marked effect on both Km and Vmax of pairs containing m6G and on the incorporation frequency of T opposite m6G, which was then only slightly favored over m6G.C. When replication was started directly opposite m6G, the kinetics appeared unaffected. These data indicate that the frequency of incorporation of C or T opposite m6G in a DNA template is dependent on the flanking neighbors and that a change of even a single base at the 3' position can have a major effect on mutagenic efficiency. Replication using Drosophila Pol alpha gave the same values for relative frequencies. Pairing of either C or T with m6G on the primer terminus did not significantly inhibit extension of the next normal base pair, in contrast to terminal mismatches of unmodified bases. It is concluded that, in the absence of repair, m6G can exhibit widely differing mutation frequencies which, in these experiments, can be as high as 85% of the replicated base. This variation in frequency of changed pairing could contribute to the occurrence of mutational "'hot spots" after replication of damaged DNA.