The molecular mechanisms for the discrepancy in outcome of initiating estrogen therapy (ET) around peri-menopause or several years after menopause in women are unknown. We hypothesize that the level of expression of a dominant negative estrogen receptor (ER) β variant, ERβ2, may be a key factor determining the effectiveness of ET in post-menopausal women. We tested this hypothesis in ovariectomized nine month-old (an age when irregular estrous cycles occur) female Sprague Dawley rats. Estradiol treatment was initiated either 6 days (Early ET, analogous to 4 months post-menopause in humans), or 180 days (Late ET, analogous to 11 years post-menopause in humans) after ovariectomy. Although ERβ2 expression increased in all OVX rats, neurogenic and neuroprotective responses to estradiol differed in Early and Late ET. Early ET reduced ERβ2 expression in both hippocampus and white blood cells, increased the hippocampal cell proliferation as assessed by Ki-67 expression, and improved mobility in the forced swim test. Late ET resulted in either no or modest effects on these parameters. There was a close correlation between the degree of ERβ2 expression and the preservation of neural effects by ET after OVX in rats, supporting the hypothesis that persistent elevated levels of ERβ2 are a molecular basis for the diminished effectiveness of ET in late post-menopausal women. The correlation between the expression of ERβ2 in circulating white blood cells and brain cells suggests that ERβ2 expression in peripheral blood cells may be an easily accessible marker to predict the effective window for ET in the brain.