Mouse embryos lacking the retinoic acid receptor RXRalpha properly undergo the early steps of heart development, but then fail to initiate a proliferative expansion of cardiomyocytes that normally results in the formation of the compact zone of the ventricular chamber wall. RXRalpha(-/-) embryos have a hypoplastic ventricular chamber and die in midgestation from cardiac insufficiency. In this study, we have investigated the underlying mechanistic basis of this phenotype. We find that interference with retinoic acid receptor function in the epicardium of transgenic embryos recapitulates the hypoplastic phenotype of RXRalpha deficient embryos. We further show that wild type primary epicardial cells, and an established epicardial cell line (EMC cells), secrete trophic protein factors into conditioned media that stimulate thymidine incorporation in primary fetal cardiomyocytes, and thymidine incorporation, cell cycle progression, and induction of cyclin D1 and E activity in NIH3T3 cells. In contrast, primary epicardial cells derived from RXRalpha(-/-) embryos and an EMC subline constitutively expressing a dominant negative receptor construct both fail to secrete activity into conditioned media. The production of trophic factors is induced by retinoic acid treatment and is inhibited by a retinoid receptor antagonist. Fetal atrial and ventricular myocytes both respond to epicardial-derived trophic signaling, although postnatal cardiomyocytes are nonresponsive. We therefore propose that the fetal epicardium, in response to retinoic acid and in a manner requiring the activity of RXRalpha, secretes trophic factors which drive fetal cardiomyocyte proliferation and promote ventricular chamber morphogenesis.