We propose an interhelical salt bridge rule to explain the dimerization specificity between the two amphipathic alpha-helices in the leucine zipper structure. Using the bZIP class of DNA-binding proteins as a model system, we predicted and designed novel dimerization partners. We predicted that ATF4, a member of the ATF/CREB family of transcription factors, would preferentially form heterodimers with IGEBP1, a member of the C/EBP superfamily. These predictions were verified using a gel mobility-shift assay. To further test the value of this interhelical salt bridge rule, we modified the bZIP protein C/EBP attempting to design molecules that would form preferentially heterodimers with C/EBP or molecules that would not interact with C/EBP. These designed molecules behaved as predicted. Therefore, we conclude that this interhelical salt bridge rule is useful in understanding the dimerization specificity of bZIP proteins. In addition, we suggest that this rule could be used to design novel "dominant-negative" molecules to specifically inhibit the function of target leucine zipper proteins in vivo.