BACKGROUND: Hepatic adverse events associated with the use of nonaspirin drugs and NSAIDs are uncommon, but the widespread use of these drugs may impact public health. OBJECTIVE: We conducted a case/noncase analysis of spontaneous reports to compare the hepatic safety profile of cyclooxygenase (COX)-2 selective inhibitors with that of nonselective NSAIDs. METHODS: This case/noncase analysis was conducted using the US Food and Drug Administration Freedom of Information (FDA/FOI) database (through quarter 1, 2003) and the World Health Organization Uppsala Monitoring Centre (WHO/UMC) database (through quarter 3, 2003). Council for International Organizations of Medical Sciences and WHO Adverse Reaction Terminology preferred terms were used to classify hepatic disorders with broad and specific case definitions. After reports involving established hepatotoxic drugs (bromfenac, nimesulide, sulindac) were excluded, the proportion of reports (PRs) of each case definition was calculated for each NSAID. Crude and adjusted reporting odds ratios (RORs) were used to compare the overall proportions of hepatic disorders and hepatic failure of celecoxib and rofecoxib versus nonselective NSAIDs. RESULTS: A total of 158,539 and 185,253 reports of NSAIDs were identified in the FDA/FOI and WHO/UMC databases and 25% and 16%, respectively, involved other hepatotoxic drugs. The PRs of hepatic disorders for all COX-2 selective inhibitors and non-selective NSAIDs were 3.0% in the FDA/FOI database and 2.7% in the WHO/UMC database. In the FDA/FOI and WHO/UMC databases, respectively, mmesulide (16.7% and 14.4%), bromfenac (12.0% and 20.7%), diclofenac (8.1% and 4.7%), and sulindac (6.1 % and 9.9%) were reported to be associated with higher proportions of overall hepatic disorders compared with those of other NSAIDs. Crude and adjusted RORs for the prevalences of overall hepatic disorders and hepatic failure with celecoxib and rofecoxib versus the other NSAIDs were <1 (indicating that the proportion was not higher than that of the comparator) in both databases. The interpretation of the results was unchanged when bromfenac, nimesulide, and sulindac were excluded from the analysis. CONCLUSIONS: In this case/noncase analysis, bromfenac, nimesulide, sulindac, and diclofenac had higher proportions of reports of hepatic disorders compared with those of other NSAIDs in the FDA/FOI and WHO/UMC databases. The analysis did not raise a safety concern for celecoxib or rofecoxib versus NSAIDs for overall hepatic disorders and hepatic failure.