The hydrogen bonding interactions between the Klenow fragment of Escherichia coli DNA polymerase I with the proofreading exonuclease inactivated (KF(-)) and the minor groove of DNA were examined with modified oligodeoxynucleotides in which 3-deazaguanine (3DG) replaced guanine. This substitution would prevent a hydrogen bond from forming between the polymerase and that one site on the DNA. If the hydrogen bonding interaction were important, then we should observe a decrease in the rate of reaction. The steady-state and pre-steady-state kinetics of DNA replication were measured with 10 different oligodeoxynucleotide duplexes in which 3DG was placed at different positions. The largest decrease in the rate of replication was observed when 3DG replaced guanine at the 3'-terminus of the primer. The effect of this substitution on mispair extension and formation was then probed. The G to 3DG substitution at the primer terminus decreased the k(pol) for the extension past G/C, G/A, and G/G base pairs but not the G/T base pair. The G to 3DG substitution at the primer terminus also decreased the formation of correct base pairs as well as incorrect base pairs. However, in all but two mispairs, the effect on correct base pairs was much greater than that of mispairs. These results indicate that the hydrogen bond between Arg668 and the minor groove of the primer terminus is important in the fidelity of both formation and extension of mispairs. These experiments support a mechanism in which Arg668 forms a hydrogen bonding fork between the minor groove of the primer terminus and the ring oxygen of the deoxyribose moiety of the incoming dNTP to align the 3'-hydroxyl group with the alpha-phosphate of the dNTP. This is one mechanism by which the polymerase can use the geometry of the base pairs to modulate the rate of formation and extension of mispairs.