INTRODUCTION: The Outcome Measures in Rheumatology (OMERACT) '3x3' method analyzes the occurrence of benefit and harm simultaneously at the individual patient level. We applied this method to two recent rheumatoid arthritis (RA) trial datasets. METHODS: Treatment of Early Aggressive RA (TEAR) and the RA Comparison of Active Therapies (RACAT) randomized trial outcomes for safety were defined according to OMERACT as no events, non-serious events, and serious events. Treatment efficacy was defined as good, moderate, or no response. A good treatment response without any events was labeled an 'unqualified success', and no treatment response but at least one adverse event an 'unmitigated failure'. Chi-square or exact tests assessed the association between benefit and harm, as appropriate. RESULTS: In TEAR, 612 of 755 patients had response data at 48 weeks: 14% of patients experienced unqualified success and 9% unmitigated failure, with no difference between the treatment arms. Treatment response and adverse event rates were not correlated. In RACAT, 309 of 353 patients had response data at 48 weeks: 6% of patients experienced unqualified success and 11% unmitigated failure, with no differences between the treatment arms. Response and adverse event rates were negatively correlated: frequency of AE and SAE increased as response decreased (p=0.008). CONCLUSION: We found some evidence that clinical response may be reduced by the co-occurrence of adverse events. This article is protected by copyright. All rights reserved.