Despite advances in the understanding of its molecular pathophysiology, pancreatic cancer remains largely incurable, highlighting the need for novel therapies. We developed a chimeric antigen receptor (CAR) specific for prostate stem cell antigen (PSCA), a glycoprotein that is overexpressed in pancreatic cancer starting at early stages of malignant transformation. To optimize the CAR design, we used antigen-recognition domains derived from mouse or human antibodies, and intracellular signaling domains containing one or two T cell costimulatory elements, in addition to CD3zeta. Comparing multiple constructs established that the CAR based on human monoclonal antibody Ha1-4.117 had the greatest reactivity in vitro. To further analyze this CAR, we developed a human pancreatic cancer xenograft model and adoptively transferred CAR-engineered T cells into animals with established tumors. CAR-engineered human lymphocytes induced significant antitumor activity, and unlike what has been described for other CARs, a second-generation CAR (containing CD28 cosignaling domain) induced a more potent antitumor effect than a third-generation CAR (containing CD28 and 41BB cosignaling domains). While our results provide evidence to support PSCA as a target antigen for CAR-based immunotherapy of pancreatic cancer, the expression of PSCA on selected normal tissues could be a source of limiting toxicity.