Human progesterone receptors (hPR) exist as two isoforms: 120 kDa B-receptors (hPR(B)) and N-terminally truncated 94 kDa A-receptors (hPR(A)). When transfected separately, each isoform exhibits different transcriptional properties that are ligand- and promoter-specific. In human target tissues, both receptor isoforms are present, so that a mixture of three dimeric species, A/A, A/B, and B/B, bind to DNA at progesterone response elements (PRE), and regulate transcription. To study the transcriptional phenotype of pure A/B heterodimers uncontaminated by A/A or B/B homodimers, we exploited the property of the leucine zipper (zip) domains of fos and jun, to form pure heterodimers. Chimeric constructs were made linking the zip of either c-fos or c-jun to the C-terminus of hPR(B) or hPR(A) (hPR-zip) to produce A-fos, B-fos, A-jun or B-jun. To determine whether the A- or B-isoform is functionally dominant in the A/B heterodimer, cells expressing hPR-zip chimeras were treated with the progestin antagonist RU486, which produces opposite transcriptional effects with the two isoforms. Gel mobility shift and immune co-precipitation assays show that in the presence of RU486 only pure heterodimers form between A-fos/B-jun or A-jun/B-fos, and bind DNA at PREs. Thus, in these pairs, interactions between the extrinsic fos/jun zipper domains override interactions between the intrinsic hPR dimerization domains. We find that under these conditions, antagonist-occupied B-zip homodimers stimulate transcription, while antagonist-occupied A-zip homodimers are inhibitory, and that pure A/B zip heterodimers have the inhibitory transcriptional phenotype of the A-zip homodimers. We conclude that, in pure heterodimers, A-receptors are dominant negative inhibitors of B-receptors. Additionally, the pure PR-zip heterodimers, unlike wild-type receptors, bind a PRE in the absence of hormone but do not activate transcription. Thus, PR dimerization and PRE binding are necessary but, without hormone, not sufficient to activate transcription.