The effects of estrogen and estrogen agonists can be mediated by estrogen receptor α (ERα) and estrogen receptor β (ERβ). We now report the identification and initial characterization of several novel isoforms of rat ERβ messenger RNA (mRNA). The most abundant of these mRNA variants we have called ERβ2. ERβ2 had an in-frame insertion of 54 nucleotides that resulted in the predicted insertion of 18 amino acids within the ligand binding domain. We demonstrated by semiquantitative RT-PCR and RNase protection that ERβ2 mRNA was expressed at levels equal to those of the previously published ERβ (ERβ1) in ovary, prostate, pituitary, and muscle. In tissues of the nervous system, including frontal cortex, hippocampus, and hypothalamus, ERβ1 was present in a 2- to 6-fold greater abundance than ERβ2. We have also detected variants of both ERβ1 and ERβ2 mRNAs that contained deletions of 117 bp encompassing the region encoding the second zinc finger of the DNA binding domain. All four mRNA species were efficiently translated into functional protein in a heterologous system. ERβ2 bound estradiol with a lower affinity (K_d 5.1 nm) than either ERα (0.19 nm) or ERβ1 (0.14 nm). The binding of ERβ2 was selective in that cortisol, testosterone, aldosterone, and progesterone among other agents did not compete for estradiol binding. However, a variety of known estrogenic agents, including physiological estrogens (estrone and estriol), plant and environmental estrogens (genistein, coumestrol, bisphenol A, methoxychlor), and pharmocological agents (tamoxifen, 4-hydroxytamoxifen) did effectively compete for estradiol binding to both ERβ1 and ERβ2. Interestingly, the binding pharmacology differed among the agents tested. For example, genistein competed effectively for estradiol binding to ERβ1 but was >150-fold weaker at competing from ERβ2. In contrast, 4-hydroxytamoxifen competed equally well at both receptors. We have also demonstrated by a gel shift assay that both ERβ1 and ERβ2 bound specifically to DNA containing a consensus estrogen response element. ERβ1 and ERβ2 could heterodimerize with each other and with ERα. Both ERβ1 and ERβ2 activated transcription in response to estradiol, however, ERβ2 required a 1000-fold greater estradiol concentration for activity than did ERβ1. Cotransfection of ERβ2 had no effect on ERβ1 activation when used in a equal ratio. A 10-fold excess of ERβ2 did raise the half-maximal dose of estradiol required for transcriptional activation, whereas the maximal level of induction did not change. The ERβ complementary DNAs deleted within the DNA binding domain could not bind to DNA or activate transcription from this reporter in the cell backgrounds tested. In conclusion, although the physiological significance of these ERβ variants warrants further investigation, ERβ2 mRNA encodes a specific, functional receptor for estradiol and estrogenic agents. We propose that ERβ2 should also be considered in addition to ERβ1 and ERα when describing the effects of estrogen, estrogen agonists/antagonists, or environmental estrogens.