Cyclooxygenase (COX) inhibition has been shown to prevent the development of esophageal adenocarcinoma (EAC). However, the potential of this approach for treatment of established cancer has been poorly investigated. Our objective was to determine whether non-selective or selective inhibition of the COX pathway affects the growth of esophageal adenocarcinoma xenografts in nude mice. A human esophageal adenocarcinoma xenograft model was established by subcutaneous inoculation of OE33 cells in nude mice. Small tumor slices harvested from four OE33 xenografts were implanted in the flanks of new mice that were randomized to different treatments (6 animals per group): indomethacin (3 mg/kg/day), parecoxib (0.11 and 0.22 mg/kg/day) or a selective prostaglandin E₂ receptor antagonist (AH-23848B, 1 mg/kg/day). For each treatment, a control group of 6 animals (vehicle) carrying xenografts from the same OE33 tumor was included. Tumor growth was measured twice a week. After 8 weeks mice were euthanized. Tumors were assessed by histological analysis, mRNA expression of COX isoenzymes, PGE₂ receptors and PGE₂ content. All OE33 tumors were poorly differentiated esophageal adenocarcinomas. Tumors expressed COX-2, EP₁, EP₂ and EP₄ receptor mRNA. Treatment with parecoxib, higher dose or indomethacin significantly inhibited tumor growth. Furthermore, indomethacin induced tumor regression (74 vs 582% in control animals; p<0.01). However, AH-23848B or parecoxib low dose failed to affect tumor growth significantly. PGE₂ content in tumors was significantly decreased by high-dose parecoxib and indomethacin. Indomethacin and parecoxib inhibit the growth of human esophageal adenocarcinoma xenografts in nude mice, which suggests a potential role for NSAIDs or selective COX-2 inhibitors for EAC chemotherapy.