Estradiol-17 (E 2 ) acts through the estrogen receptor (ER) to regulate uterine growth and functional differentiation. To determine whether E 2 elicits epithelial mitogenesis through epithelial ER versus indirectly via ERpositive stromal cells, uteri from adult ER-deficient ER knockout (ko) mice and neonatal ER-positive wild-type (wt) BALB͞c mice were used to produce the following tissue recombinants containing ER in epithelium (E) and͞or stroma (S), or lacking ER altogether: wt-S ؉ wt-E, wt-S ؉ ko-E, ko-S ؉ ko-E, and ko-S ؉ wt-E. Tissue recombinants were grown for 4 weeks as subrenal capsule grafts in intact female nude mice, then the hosts were treated with either E 2 or oil a week after ovariectomy. Epithelial labeling index and ER expression were determined by [ 3 H]thymidine autoradiography and immunohistochemistry, respectively. In tissue recombinants containing wt-S (wt-S ؉ wt-E, wt-S ؉ ko-E), E 2 induced a similar large increase in epithelial labeling index compared with oil-treated controls in both types of tissue recombinants despite the absence of epithelial ER in wt-S ؉ ko-E tissue recombinants. This proliferative effect was blocked by an ER antagonist, indicating it was mediated through ER. In contrast, in tissue recombinants prepared with ko-S (ko-S ؉ ko-E and ko-S ؉ wt-E), epithelial labeling index was low and not stimulated by E 2 despite epithelial ER expression in ko-S ؉ wt-E grafts. In conclusion, these data demonstrate that epithelial ER is neither necessary nor sufficient for E 2 -induced uterine epithelial proliferation. Instead, E 2 induction of epithelial proliferation appears to be a paracrine event mediated by ER-positive stroma. These data in the uterus and similar studies in the prostate suggest that epithelial mitogenesis in both estrogen and androgen target organs are stromally mediated events.