High-throughput methods are powerful tools to develop predictive models for assessing drug-induced liver injury (DILI). However, the development of predictive models requires a drug reference list with an accurate annotation of DILI risk in humans. We previously developed a DILI annotation schema based on information curated from the US Food and Drug Administration (FDA)-approved drug labeling for 287 drugs. In this article, we refine the schema by weighing the evidence of causality (i.e., a verification process to evaluate a drug as the cause of DILI, using resources such as clinical causality assessments including the Roussel Uclaf Causality Assessment Method (RUCAM), Drug-Induced Liver Injury Network (DILIN) causality committee, and National Institutes of Health (NIH) LiverTox database) and generate a data set (DILIrank) that ranks the DILI risk in humans for 1036 FDA-approved drugs. This data set classifies drugs into four categories: verified Most-DILI concern (vMost-DILI concern, 192 drugs), verified Less-DILI concern (vLess-DILI concern, 278 drugs), verified No-DILI concern (vNo-DILI concern, 312 drugs), and Ambiguous DILI concern (254 drugs), providing the largest annotated data set of such drugs in the public domain. We analyzed the DILI landscape using the DILIrank data set based on therapeutic categories (e.g., nonsteroidal anti-inflammatory drugs (NSAIDs), antivirals) and chemical subgroups (e.g., nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs), protein kinase inhibitors), identifying those enriched with higher DILI risk. This refined schema and DILIrank data set improve the accuracy of DILI risk annotation, supporting the development of predictive models and biomarkers for DILI assessment in the era of data-driven science.