Pancreatic ductal adenocarcinomas are thought to arise from noninvasive, intraductal precursor lesions called pancreatic intraepithelial neoplasias (PanIN). The study of PanINs holds great promise for the identification of early detection markers and effective cancer-preventing strategies. Cyclooxygenase-2 (COX-2) represents an intriguing target for therapeutic and preventive approaches in various human malignancies. The aim of the present study was to evaluate the efficacy of a selective COX-2 inhibitor to prevent the progression of PanINs in a conditional Kras^G12D mouse model. Offspring of LSL-KRAS^G12D x PDX-1-Cre intercrosses were randomly allocated to a diet supplemented with the selective COX-2 inhibitor nimesulide (400 ppm) or a control diet. After 10 months, animals were sacrificed. Successful recombination in the pancreas was evaluated by PCR. The pancreas of KRAS^G12D;PDX-1-Cre mice was analyzed for the presence of murine PanINs. Animals fed the COX-2 inhibitor had significantly fewer PanIN-2 and PanIN-3 lesions than control animals (P < 0.05). Ten percent of all pancreatic ducts in the nimesulide-fed animals showed PanIN-2 or PanIN-3 lesions, whereas 40% of the pancreatic ducts in the control animals had PanIN-2 or PanIN-3 lesions. Intrapancreatic prostaglandin E₂ levels were reduced in nimesulide-fed animals. Immunohistochemistry confirmed COX-2 expression in early and late PanINs. In summary, we found that the selective COX-2 inhibitor nimesulide delays the progression of pancreatic cancer precursor lesions in a preclinical animal model. These data highlight the importance of COX-2 in the development of pancreatic cancer. Inhibition of COX-2 may represent an intriguing strategy to prevent pancreatic cancer in high-risk patients. [Cancer Res 2007;67(15):7068–71]