Nonsteroidal anti-inflammatory drug (NSAID) gastropathy is sufficiently important as to warrant co-administration of misoprostol or proton pump inhibitors or a switch to selective cyclooxygenase (COX)-2 inhibitors. However, the serious ulcer outcome studies suggested that 40% of the clinically significant gastrointestinal bleeding originated more distally, presumably from NSAID enteropathy. We used capsule enteroscopy to study small-bowel damage in patients on long-term NSAIDs and COX-2-selective agents. Sixty healthy volunteers acted as controls. One hundred twenty and 40 patients on long-term NSAIDs and COX-2 selective agents, respectively, underwent a capsule enteroscopy study. Small-bowel damage was categorized and quantitated. Sixty-two percent of patients on conventional NSAIDs were abnormal, which differed significantly (P < .001) from controls. The main pathology related to reddened folds (13%), denuded areas (39%), and mucosal breaks (29%). Two percent had diaphragm-like strictures and 3% had bleeding without an identifiable lesion. The damage, seen in 50% of patients on selective COX-2 inhibitors (reddened folds, 8%; denuded areas, 18%; and mucosal breaks, 22%), did not differ significantly (P > .5) from that seen with NSAIDs. Long-term NSAIDs and COX-2-selective agents cause comparable small-bowel damage. This suggests an important role for COX-2 in the maintenance of small-bowel integrity. The results have implications for strategies that aim to minimize the gastrointestinal damage in patients requiring anti-inflammatory analgesics.