More effective therapeutic approaches for castration-resistant prostate cancer (CRPC) are urgently needed, thus reinforcing the need to understand how prostate tumors progress to castration resistance. We have established a novel mouse xenograft model of prostate cancer, KUCaP-2, which expresses the wild-type androgen receptor (AR) and which produces the prostate-specific antigen (PSA). In this model, tumors regress soon after castration, but then reproducibly restore their ability to proliferate after 1 to 2 months without AR mutation, mimicking the clinical behavior of CRPC. In the present study, we used this model to identify novel therapeutic targets for CRPC. Evaluating tumor tissues at various stages by gene expression profiling, we discovered that the prostaglandin E receptor EP4 subtype (EP4) was significantly upregulated during progression to castration resistance. Immunohistochemical results of human prostate cancer tissues confirmed that EP4 expression was higher in CRPC compared with hormone-naïve prostate cancer. Ectopic overexpression of EP4 in LNCaP cells (LNCaP-EP4 cells) drove proliferation and PSA production in the absence of androgen supplementation in vitro and in vivo. Androgen-independent proliferation of LNCaP-EP4 cells was suppressed when AR expression was attenuated by RNA interference. Treatment of LNCaP-EP4 cells with a specific EP4 antagonist, ONO-AE3-208, decreased intracellular cyclic AMP levels, suppressed PSA production in vitro, and inhibited castration-resistant growth of LNCaP-EP4 or KUCaP-2 tumors in vivo. Our findings reveal that EP4 overexpression, via AR activation, supports an important mechanism for castration-resistant progression of prostate cancer. Furthermore, they prompt further evaluation of EP4 antagonists as a novel therapeutic modality to treat CRPC.