Many studies have shown that a variety of strategies, including the use of cyclooxygenase-2 (COX-2) inhibitors, or co-prescription of misoprostol or proton pump inhibitors, result in reduced endoscopic damage and ulceration compared with non-selective nonsteroidal anti-inflammatory drugs (NSAIDs) alone. Questions have been raised as to whether this would translate into improved clinical outcomes. Consequently, several studies have investigated whether use of COX-2 inhibitors (Vioxx Gastrointestinal Outcomes Research [VIGOR] and the Celecoxib Long-Term Arthritis Safety Assessment Study [CLASS] studies) or co-prescription with misoprostol (MUCOSA) would reduce the event rate of clinically significant ulcers. These studies have shown an approximate halving of such events. They have raised the possibility that use of low dose aspirin may compromise these benefits and appear to have shown differences between (at least some) COX-2 inhibitors and (at least some) NSAIDs with regard to myocardial infarction. Among the lessons learned from this experience are the need to define closely the outcomes of interest and possibly to concentrate on ulcer complications, the need for adequately powered studies, and the fact that endoscopic studies broadly predict outcomes. However, there are differences in the estimated rates of reduction. It is not self evident whether outcomes studies or endoscopic studies give a truer estimate of risk. A helpful development would be more standardized gastrointestinal assessment at the time of ulcer complications and this could be achieved if studies were done in countries with well-developed primary care systems.