Background The VIOXX Gastrointestinal Outcomes Research (VIGOR) trial showed a 53% decrease in the risk of upper gastrointestinal toxicity and a fivefold increase in the risk of myocardial infarction for rofecoxib (a selective cyclooxygenase-2 inhibitor) compared with naproxen. We examined the effects of these competing adverse events on life expectancy in patients with rheumatoid arthritis. Methods We used decision analysis to compare the life expectancy of a cohort of rheumatoid arthritis patients taking naproxen versus a similar cohort taking rofecoxib, using data from the VIGOR trial. We incorporated the competing risks of upper gastrointestinal toxicity and myocardial infarction, as well as their long-term health consequences, on the basis of population-based studies. Results For 58-year-old women with rheumatoid arthritis (i.e., typical of participants in the VIGOR trial), naproxen was associated with a longer life expectancy than was rofecoxib (difference = 4.4 months). This difference was larger among 58-year-old men (7.8 months). The probability that naproxen is associated with a longer life expectancy than rofecoxib among 58-year-old patients was 92% for women and 98% for men. Life expectancy became the same between the two treatments when the risk of upper gastrointestinal toxicity was 70% higher or the risk of myocardial infarction was 40% lower than that of the base case among women, and when the risk of upper gastrointestinal toxicity was 4.4-fold higher or the risk of myocardial infarction was 70% lower among men. Conclusion Our analysis suggests that the competing risks of upper gastrointestinal toxicity and myocardial infarction shown in the VIGOR trial would project a longer life expectancy with naproxen than rofecoxib among patients with rheumatoid arthritis, except in those at low risk of myocardial infarction or at high risk of upper gastrointestinal toxicity.