Objective: To determine whether audit/feedback and educational materials improve adherence to recommendations for laboratory monitoring and cytoprotective agents to detect and prevent adverse events caused by nonsteroidal anti-inflammatory agents (NSAIDs). Study Design: Controlled, cluster-randomized trial. Methods: Physicians commonly prescribing NSAIDs were identified within a large managed care organization and randomized to a control or an intervention group (audit/feedback with peer-derived benchmarks and continuing medical education). Medical records were examined 10 months before and after the intervention for clinical data and receipt of complete blood count (CBC), creatinine testing, and cytoprotective agents (process measures). Primary analysis compared intervention versus control physicians among those who initially performed below a peer-derived benchmark. General estimating equations accounted for patient clustering. Results: Of 101 physicians initially randomized, 85 remained eligible (38 internists, 36 family physicians, 11 rheumatologists) postintervention. Mean percent change in performance between intervention and control physicians for CBC monitoring was 16% versus 10%; for creatinine monitoring, 0% versus 17%; and use of cytoprotective agents, -3% versus -1%. None of these changes were significant. Rheumatology specialty, number of NSAID prescriptions and physician visits, and patient risk factors for NSAID-related toxicity were more strongly associated with improved safety practices than the intervention. Conclusions: Audit/feedback and educational materials had no observed effect on improving NSAID-related safety practices. Potentially contributing factors include high baseline performance (ceiling effect), dilution of the intervention effect by case mix and provider factors, nonreceipt of intervention materials, and diverse indications for lab tests.