Even the first expression profiling studies of breast cancers have generated new insights. They suggest for example, that information about tumor aggressiveness, prognosis, metastatic potential, or treatment outcome is encoded in, and can be deduced from, the primary tumor. On the other hand no clinical genomic array data have yet been published that deal with hormonal aspects of breast tumorigenesis, tumor progression, or therapeutics. Rather, studies have focused on experimental model systems. We review below the currently published data on array profiling in clinical breast cancer, then describe our studies in breast cancer cell lines and xenograft models dealing with progesterone receptors (PRs) and the role of progesterone. We demonstrate that the two PR isoforms, PR-A and PR-B, have mostly nonoverlapping molecular signatures when liganded by progesterone, with PR-B the more active form. Additionally, we document the surprising finding that unliganded PRs can regulate gene transcription, with PR-A the more active form. In ovariectomized mice supplemented with estradiol but lacking measurable progesterone, PR-B-expressing tumors grow to twice the size of PR-A-expressing ones. We conclude that in breast cancers, PRs are more than simple markers of estrogen receptor function. Rather, presence of PRs and the ratio of the two isoforms directly influence tumor phenotype, even in the absence of ligand.