Objective: Present results from two hepatic safety studies conducted within 20 months after duloxetine launch. Methods: Signal detection based on spontaneous reports to the FDA adverse event reporting system (AERS) and on a comparison of duloxetine and venlafaxine in the i3 Drug Safety Aperio claims database, using measures of disproportionality and incidence rate ratio, respectively. Results: In AERS all antidepressants had some degree of association with hepatic injury, in that at least one hepatic event was disproportionately represented for each drug. Signals were detected for duloxetine cases analyzed against full and antidepressant- only backgrounds. These signals corresponded to labeled events or events investigated during ongoing surveillance. Using a duloxetine fatal-case series, disproportional representation of clinically serious events was detected relative to both backgrounds, but the signals were refuted upon independent expert panel case review. The Aperio study showed no difference in hepatic injury between duloxetine and venlafaxine initiators after proper control for baseline risks, suggesting differential prescribing of duloxetine, perhaps preferentially as second-line therapy in some initiators. Conclusions: No new signals were identified in Aperio. New signals detected through AERS were refuted upon independently conducted case-level investigation. Hepatic signals arising from spontaneously reported data must be clarified through subsequent systematic investigation.