Human esophageal adenocarcinoma (EAC) develops in a sequence from gastroesophageal reflux disease (GERD), columnar-lined esophagus (CLE), dysplasia, and eventually to EAC. We established a rat surgical EAC model with esophagogastroduodenal anastomosis (EGDA) to mimic the staged process of esophageal adenocarcinogenesis. Profiling of the AA metabolites with mass spectrometry showed that prostaglandin E2 (PGE2), leukotriene B4 (LTB4), 15-hydroeicosatetraenoic acid (HETE), 12-HETE, 8-HETE and 5-HETE all increased at the esophagoduodenal junction after EGDA as compared with the proximal esophagus, with PGE2 as the major metabolite. Consistent with this profile, cyclooxygenase 2 (Cox2) was overexpressed in the basal cell layer of esophageal squamous epithelium, CLE cells and EAC tumor cells of the EGDA rats, as compared with the normal esophageal epithelium. Sulindac (a Cox inhibitor), nordihydroguaiaretic acid (NDGA, a lipoxygenase inhibitor) and alpha-difluoromethylornithine (DFMO, an ornithine decarboxylase inhibitor) were tested for their possible inhibitory actions against the formation of EAC in the rat EGDA model. In a short-term study (for 4 weeks after surgery), dietary administration of both sulindac (300 and 600 p.p.m.) and NDGA (100 p.p.m.) effectively reduced the EGDA-induced inflammation. In a long-term chemoprevention study (for 40 weeks after surgery), 300 p.p.m. sulindac, alone or in combination with 100 p.p.m. NDGA or 0.5% DFMO, decreased the tumor incidence from 57.7 to 26.9%, or 16.7 or 20%, respectively (P < 0.05). NDGA alone (100 and 200 p.p.m.) slightly decreased the tumor incidence to 52.4 and 37%, respectively, although the difference was not statistically significant. DFMO alone did not show significant effects on tumor incidence. Inhibition of tumor formation by sulindac was correlated with lowered levels of PGE2. In conclusion, sulindac exerted its chemopreventive effect against the formation of EAC in the rat EGDA model possibly through its inhibition of Cox.