Mixed reflux of the gastroduodenal contents induces the esophageal mucosal damage and inflammation progressing chronic esophagitis and premalignant Barrett's esophagus (BE). The role of cyclooxygenase-2 (COX-2) and chronic inflammation in the progression of BE toward adenocarcinoma of the esophagus has not been extensively studied in experimental models of BE in animals and in human subjects. We evaluated the expression of COX-2 in rat model of BE and examined the usefulness of COX-2 expression in determining the risk of malignant transformation in patients with BE treated with argon plasma coagulation (APC) that allows for effective ablation of metaplastic mucosa (group A) without or with proton pump inhibitors (PPI). In addition, the group B of patients was subjected to laparoscopic Nissen's fundoplication and group K that served as control, received PPI treatment only. Expression of COX-2 was evaluated in fresh-frozen biopsy specimens obtained from the distal esophagus in all 60 patients before and 12 months after treatment. In experimental studies, eighty rats were surgically prepared with esophagogastroduodenal anastomosis (EGDA) resulting in chronic esophagitis. At 4 months, the esophageal damage in EGDA rats was evaluated by macroscopic and histological index score, the plasma IL-1beta and TNF-alpha levels was determined by ELISA and the mucosal expression of COX-2 mRNA and COX-2 protein were assessed by RT-PCR and Western Blot, respectively. Chronic esophagitis was developed in all EGDA animals followed by the rise in the plasma TNF-alpha and IL-1beta levels. Histology revealed extensive esophageal ulcerations with development of columnar epithelium, formation of mucus glands in squamous epithelium, intestinal metaplasia distant to anastomosis consisting of goblet cells, infiltration of inflammatory cells including plasma cells and lymphocytes. COX-2 mRNA was absent in the esophageal mucosa of sham-control animals but strongly upregulated in metaplastic Barrett's epithelium. In BE patients, the overexpression of COX-2 was documented in patients with dysplasia. After APC (group A) or Nissen's fundoplication (group B), the expression of COX-2 mRNA was markedly reduced and these effects were positively correlated with histopathological findings. Controls failed to show significant alterations in COX-2 expression. We conclude that 1) EGDA rats serve as the suitable model of the chronic esophagitis by the gastrointestinal refluxate resembling many features of those observed in human Barrett's esophagus, as confirmed by severe morphology changes, excessive release of proinflammatory cytokines TNF-alpha and IL-1beta and overexpression of COX-2, and 2) the significant correlation of the degree of COX-2 overexpression with histopathological findings indicates the usefulness of this inducible biomarker as a valuable indicator of the risk of malignant transformation in patients with BE.