We discovered that pigment epithelium-derived factor (PEDF)-null mice have endometrial hyperplasia, the precursor to human type I endometrial cancer (ECA), which is etiologically linked to unopposed estrogen (E2), suggesting that this potent antiangiogenic factor might contribute to dysregulated growth and the development of type I ECA. Treatment of both ECA cell lines and primary ECA cells with recombinant PEDF dose dependently decreased cellular proliferation via an autocrine mechanism by blocking cells in G1 and G2 phases of the cell cycle. Consistent with the known opposing effects of E2 and progesterone (Pg) on endometrial proliferation, Pg increases PEDF protein synthesis and release, whereas E2 has the converse effect. Using PEDF luciferase promoter constructs containing two Pg and one E2 response elements, E2 reduced and Pg increased promoter activity due to distal response elements. Furthermore, E2 decreases and Pg increases PEDF secretion into conditioned media (CM) by both normal endometrial stromal fibroblasts (ESFs) and cancer-associated fibroblasts (CAFs), but only CM from ESFs mediated growth-inhibitory activity of primary endometrial epithelial cells (EECs). In addition, in cocultures with primary EECs, Pg-induced growth inhibition is mediated by ESFs, but not CAFs. This is consistent with reduced levels of Pg receptors on CAFs surrounding human malignant glands in vivo. Taken together, the data suggest that PEDF is a hormone-regulated negative autocrine mediator of endometrial proliferation, and that paracrine growth inhibition by soluble factors, possibly PEDF, released by ESFs in response to Pg, but not CAFs, exemplifies a tumor microenvironment that contributes to the pathogenesis of ECA.