We have reviewed the effects of nonsteroidal antiinflammatory drugs (NSAIDs) on the gastroduodenal mucosa and then, by means of meta-analysis, have evaluated the results of therapeutical trials in the prevention of NSAID-induced gastroduodenal mucosal damage. We searched the literature through Medline (1980-1990) and through the references of relevant articles. Of 19 trials retrieved by these means, 11 proved eligible for meta-analysis on the basis of eight selection criteria defined a priori. The data included seven studies where the effect of preventive treatment during short-term use of NSAIDs was analyzed and four studies dealing with prevention of mucosal damage in subjects treated with NSAIDs for long periods. Results were expressed in terms of the percentage of patients developing severe mucosal damage or an ulcer during short- and long-term treatment with NSAIDs, respectively. The pooling of the results showed that, during short-term NSAID use, 37% of the subjects developed severe gastric mucosal damage as compared to 12% of subjects given some protective agent. The figures for the duodenum are 13% and 4%, respectively. Owing to the small number of studies on prevention of chronic NSAID-induced gastroduodenal damage, results were not pooled together; misoprostol was shown to be highly effective in reducing the prevalence of gastric ulcer, and ranitidine prevented the occurrence of duodenal but not gastric ulcer. Despite these positive results, there is no proof that protective agents should be recommended to the general population requiring NSAIDs therapy. Nor is there yet evidence that taking a protective agent will avoid the complications of NSAIDs, such as bleeding or perforation.