No consensus exists as to whether cotherapy is effective in the short-term prevention of severe NSAID-related gastroduodenal damage. The aim of this study was to provide a quantitative systematic review of the efficacy of gastroprotective drugs, such as misoprostol, H2-blockers, and proton pump inhibitors (PPI) in preventing the severe acute NSAID-related gastroduodenal damage. Placebo-controlled randomized clinical trials on the use of misoprostol, H2-blockers, and PPIs as preventative agents published between January 1986 and May 1999 were identified through Medline and reference lists from clinical reviews. Studies on patients or healthy subjects were considered to be eligible for data pooling if they were performed in acute NSAID users (not longer than 30 days) and with at least one endoscopic evaluation during therapy that reported results specifically for gastric and duodenal damage. Risk difference (RD), heterogeneity chi2 test, publication bias assessment and number needed to treat (NnT) were calculated for each meta-analysis by a customized program. Twenty-one trials met the inclusion criteria evaluating a total of 636 healthy subjects and 1904 patients with arthritis randomized to active drug or placebo. The baseline risk of NSAID-related gastric (68% vs 16.6%, P < 0.001) and duodenal (22% vs 8.5%, P < 0.001) damage was higher in healthy subjects compared to patients with arthritis. Meta-analysis demonstrated a significant heterogeneity between trials performed in the two populations (P < 0.0001). In healthy subjects the active drug treatment induced a significant prevention of severe gastric (misoprostol RD = 69%, 95% CI = 60.3-77.7, H2-blocker RD = 38.3%, 95% CI = 17.8-58.9 and PPI RD = 43%, 95% CI = 28.2-57.7) and duodenal damage (misoprostol RD = 22.3%, 95% CI = 13.6-31, H2-blocker RD = 13.2%, 95% CI = 5.2-21.3 and PPI RD = 17.7%, 95% CI = 3.5-31.8). NnT values were, respectively, 1, 3, and 2 for gastric and 4, 8, and 6 for duodenal damage. In patients with arthritis lower RD and higher NnT values were found compared to healthy subjects. In conclusions, cotreatment with gastroprotective drugs for short-term prevention of severe gastroduodenal NSAID-related damage was more effective in healthy subjects than in patients with arthritis; misoprostol and PPIs were more effective than H2-blockers in the prevention of both gastric and duodenal severe damage; more studies need to evaluate the role of short-term prevention in patients with arthritis who require acute NSAID treatment.