Epidemiological studies suggest that the use of NSAIDs and/or a high intake of fruit and vegetables reduce the risk of oesophageal adenocarcinoma. Since COX-2 is up-regulated in Barrett's oesophageal carcinogenesis, the protective effect of NSAIDs and natural food components might reflect COX-2 inhibition. We explored the effects of quercetin, a natural flavonoid with a potent COX-2 inhibitory activity, and two commercially available selective COX-2 inhibitors (NS-398 and nimesulide) on cell proliferation, apoptosis, PGE2 production and COX-2 mRNA expression in a human oesophageal adenocarcinoma cell line (OE33). Changes in the relative numbers of adherent and floating cells were quantified and apoptotic cells were identified using ethidium bromide and acridine orange staining under fluorescence microscopy. Flow cytometric analysis of adherent and floating cells was used to quantify apoptosis and to examine the effects of the agents on the cell cycle. After 48 h exposure at concentrations of > or =1 microM both COX-2 inhibitors and quercetin suppressed cell proliferation (P < 0.01) and increased the fraction of floating apoptotic cells. At higher concentrations (50 microM) and longer exposure (48 h) the effects of quercetin were significantly greater than those of the selective COX-2 inhibitors (P < 0.01). Cell cycle analyses showed that quercetin blocked cells in S phase, while the selective COX-2 inhibitors blocked cells in G1/S interphase. COX-2 mRNA expression was suppressed by quercetin and the synthetic COX-2 inhibitors in a time- and dose-dependent manner. Quercetin and the synthetic COX-2 inhibitors (10 microM) suppressed PGE2 production by approximately 70% after 24 h exposure (P < 0.001). We conclude that OE33 is a useful model for the study of COX-2 expression and associated phenomena in human adenocarcinoma cells. Synthetic COX-2 inhibitors and the food-borne flavonoid quercetin suppress proliferation, induce apoptosis and cell cycle block in human oesophageal adenocarcinoma cells in vitro, and future studies should assess their effects in vivo.