The sequence of the bacteriophage 434 OR1 (ACAAAACTTTCTTGT) differs from its OR3 (ACAGTTTTCTTGT) at positions 4-6. X-ray analysis shows that the side chain of Gln33 of the 434 repressor makes van der Waals' and H-bond contacts with the T at position 4' in complex with OR1, but no specific contact is observed at this position in 434 repressor-OR3 complexes. No contacts are made by repressor to the bases at positions 5 or 6 in either binding site. The significance of the sequence differences between OR1 and OR3 in determining the operator affinity for repressor were examined by constructing synthetic variants of these operators. Measurements of the affinity of these operators for repressor as a function of ionic strength revealed that although base pairs 5 and 6 are not contacted by 434 repressor, they can nonetheless influence operator affinity for repressor by modulating the degree to which ionic interactions contribute to the overall binding energy. Both the magnitude and direction of their effect depends on the status of repressor's contacts to the bases at position 4. The role of contact made by Gln33 to position 4 was examined by mutating this amino acid to Ala and by examining the affinity of wild type repressor for an operator bearing a 5-methylcytosine at position 4' in an OR1-4G mutant. These experiments showed that repressor's preferences at operator positions 5 and 6 are linked to its position 4 preference via a van der Waals' contact between amino acid 33 and a methyl group on the base at operator position 4'. Together, the results of the experiments shown here reveal that bases that do not contact the protein alter its preferences for bases at the contacted operator position 4.