The aim of this study was to characterize voriconazole trough concentrations (Cmin) and associated hepatotoxicity, and to determine predictors of hepatotoxicity and identify high-risk groups in critically ill patients.This was a nationwide, multicenter, retrospective study. We studied Cmin and hepatotoxicity in 363 critically ill patients from 2015 to 2020 who received voriconazole treatment. Logistic regression and classification and regression tree (CART) models were used to identify the high-risk patients.A large interindividual variability was observed in the voriconazole initial Cmin and concentrations ranging from 0.1 mg/L to 18.72 mg/L. Voriconazole-related grades ≥2 hepatotoxicity developed in 101 patients, including 48 patients with grades ≥3 hepatotoxicity. The median time to hepatotoxicity occurrence was 3 days (range 1-24 days), and 83.2% of hepatotoxicity cases occurred within 7 days after voriconazole initiation. Voriconazole Cmin was significantly associated with hepatotoxicity. The CART model showed that the significant predictors of grades ≥2 hepatotoxicity were Cmin >3.42 mg/L, receiving trimethoprim-sulfamethoxazole or tigecycline treatment simultaneously, and suffering from sepsis shock. The model predicted that grades ≥2 hepatotoxicity incidence among these high-risk patients was 48.3-63.4%. The significant predictors of grades ≥3 hepatotoxicity were Cmin >6.87 mg/L, concomitating use of ≥3 hepatotoxic drugs, and complicating with sepsis shock, and the predictive incidence among these high-risk patients was 22.7-36.8%.Patients with higher voriconazole Cmin and sepsis shock, as well as concomitant use of hepatotoxic drugs were the strongest predictors of hepatotoxicity. Voriconazole plasma concentration monitoring should be performed early to avoid hepatotoxicity.Copyright © 2022. Published by Elsevier Ltd.