This study aimed to identify medications with the highest incidence rates of severe acute liver injury (ALI) using real-world data and compare these rates with those reported in case reports. The study analyzed data from the US Department of Veterans Affairs, focusing on patients without preexisting liver or biliary disease who initiated suspected hepatotoxic medications between October 1, 2000, and September 30, 2021. The primary outcome was hospitalization for severe ALI, defined by specific biochemical criteria. The study found that 17 medications had the highest observed rates of severe ALI, with rates ranging from 0 to 86.4 events per 10,000 person-years. Notably, 11 of these medications (64%) were not classified as highly hepatotoxic based on case report categories. The study concluded that real-world data can accurately identify medications with high hepatotoxicity, challenging the reliance on case reports for hepatotoxicity categorization. This approach provides a more objective method for investigating hepatotoxicity safety signals and could be applied to other electronic medical record systems.This study aimed to identify medications with the highest incidence rates of severe acute liver injury (ALI) using real-world data and compare these rates with those reported in case reports. The study analyzed data from the US Department of Veterans Affairs, focusing on patients without preexisting liver or biliary disease who initiated suspected hepatotoxic medications between October 1, 2000, and September 30, 2021. The primary outcome was hospitalization for severe ALI, defined by specific biochemical criteria. The study found that 17 medications had the highest observed rates of severe ALI, with rates ranging from 0 to 86.4 events per 10,000 person-years. Notably, 11 of these medications (64%) were not classified as highly hepatotoxic based on case report categories. The study concluded that real-world data can accurately identify medications with high hepatotoxicity, challenging the reliance on case reports for hepatotoxicity categorization. This approach provides a more objective method for investigating hepatotoxicity safety signals and could be applied to other electronic medical record systems.