The human RNA editing enzyme ADAR1 (double-stranded RNA deaminase I) deaminates adenine in pre-mRNA to yield inosine, which codes as guanine. ADAR1 has two left-handed Z-DNA binding domains, Z alpha and Z beta, at its NH(2)-terminus and preferentially binds Z-DNA, rather than B-DNA, with high binding affinity. The cocrystal structure of Z alpha(ADAR1) complexed to Z-DNA showed that one monomeric Z alpha(ADAR1) domain binds to one strand of double-stranded DNA and a second Z alpha(ADAR1) monomer binds to the opposite strand with 2-fold symmetry with respect to DNA helical axis. It remains unclear how Z alpha(ADAR1) protein specifically recognizes Z-DNA sequence in a sea of B-DNA to produce the stable Z alpha(ADAR1)-Z-DNA complex during the B-Z transition induced by Z alpha(ADAR1). In order to characterize the molecular recognition of Z-DNA by Z alpha(ADAR1), we performed circular dichroism (CD) and NMR experiments with complexes of Zalpha(ADAR1) bound to d(CGCGCG)(2) (referred to as CG6) produced at a variety of protein-to-DNA molar ratios. From this study, we identified the intermediate states of the CG6-Z alpha(ADAR1) complex and calculated their relative populations as a function of the Z alpha(ADAR1) concentration. These findings support an active B-Z transition mechanism in which the Z alpha(ADAR1) protein first binds to B-DNA and then converts it to left-handed Z-DNA, a conformation that is then stabilized by the additional binding of a second Z alpha(ADAR1) molecule.