We are using experimental and computational analyses to probe how coiled-coil-mediated interaction specificity is established in vitro, using peptides derived from the human bZIP transcription factors as a model. Beginning with comprehensive measurements of native interactions, we developed computational methods for predicting bZIP-like associations from sequence. Using these models, we have now designed synthetic peptides that bind to native human bZIP coiled-coil targets selectively. We have also tested our designed coiled-coil peptides for interactions with each other. Because the computational design process disfavored self-association, most of the synthetic peptides, which we call SYNZIPs, preferentially form hetero-oligomeric interactions. We characterized the complete set of pair-wise interactions among the SYNZIPs using microarrays and found that they can be used to build numerous heterospecific interaction motifs. Thanks to the work of many people and labs, our understanding of bZIP-like coiled-coil interactions is probably more mature than our view of any other protein interaction motif. Still, much remains to be discovered and explained.