In order to provide a systematic overview of the available information on the prevention of nonsteroidal anti-inflammatory drug (NSAID)-induced gastrointestinal mucosal injury, we performed a meta-analysis based on all the in extenso published randomized clinical trials comparing H2-blockers or misoprostol to placebo for the prevention of NSAID-induced gastrointestinal mucosal injury in arthritis patients or normal subjects. The main endpoints after NSAID therapy were considered the number of subjects developing gastric ulcer, or clinically relevant gastric lesions (i.e. more than 10 erosions or 1 ulcer), or duodenal ulcer, or clinically relevant duodenal lesions (i.e. more than 10 erosions or 1 ulcer). The total number of patients studied was 1,955, and that of normal subjects was 715. Results were analyzed by the DerSimonian and Laird method, as well as by the Peto one. The data available from the trials suggest that misoprostol prevention is of benefit in patients under NSAID treatment for the prevention of NSAID-induced gastroduodenal mucosal injury. The effect of prevention is statistically significant with both methods, and ranges from -42% (duodenal ulcer) to -79% (gastric ulcer). The prevention estimates for gastric and duodenal erosions are in between. H2-blockers seem to be less effective: prevention is not indeed demonstrable for the more relevant lesions, like gastric and duodenal ulcer.