The Leu3 protein (Leu3p) of Saccharomyces cerevisiae is a pleiotropic transregulator that can function both as an activator and as a repressor of transcription. It binds to upstream promoter elements (UASLEU) with the consensus sequence 5'-GCCGGNNCCGGC-3'. The DNA-binding motif of Leu3p belongs to the family of Zn(II)2-Cys6 clusters. The motif is located between amino acid residues 37 and 67 of the 886-residue protein. In this study, we used a recombinant peptide consisting of residues 17 to 147 to explore the interaction between Leu3p and its cognate DNA. We found that the Leu3p(17-147) peptide is a monomer in the absence of UASLEU but assumes a dimeric structure when the DNA is present. Results of protein-DNA cross-linking and methylation and ethylation interference footprinting experiments show that the Leu3p(17-147) dimer interacts symmetrically with two contact triplets separated by 6 bp and suggest that the peptide approaches its target DNA in such a way that each subunit is positioned closer to one DNA strand than to the other. The binding of Leu3p is strongly affected by the spacing between the contact triplets of the UASLEU and by the type of triplet. Binding occurs when the triplets are 6 bp apart (normal spacing) but fails to occur when the triplets are 0, 5, or 8 bp apart. Weak binding occurs when the triplets are 7 bp apart. Binding does not occur when the UASLEU triplets (GCC....GGC) are replaced with triplets found in the UAS elements for Gal4p, Put3p, and Ppr1p (CGG....CCG). The apparent Kd for the normal Leu3p(17-147)-UASLEU complex is about 3 nM. A mutant form of Leu3p(17-147) in which the histidine at position 50 has been replaced with cysteine binds UASLEU with significantly greater affinity (apparent Kd of about 0.7 nM), even though the interaction between the mutant peptide and target DNA appears to be unchanged. Interestingly, repression of basal-level transcription, which is a hallmark property of the wild-type Leu3p(17-147) peptide, is largely lost with the mutant peptide, indicating that there is no direct correlation between strength of binding and repression.