Case reports and analyses of clinical studies and of pharmacovigilance data suggest that new oral anticoagulants (NOACs) are associated with a small risk for hepatotoxicity. The objective of this publication is to summarize the current data about this subject, with a special emphasis on pharmacovigilance data in the World Health Organization (WHO) Global Individual Case Safety Reports (ICSR) database and on potential mechanisms of hepatotoxicity. For that, all available case reports as well as published analyses of clinical studies were obtained with a detailed search in PubMed. In addition, pharmacovigilance data from VigiBase(®), the WHO Global ICRS database, were extracted and analyzed. The data show that liver injury associated with NOACs was reported in clinical studies and in pharmacovigilance databases. Several case reports described potentially life-threatening hepatotoxicity in patients treated with rivaroxaban or dabigatran. For rivaroxaban, most affected patients were symptomatic and liver injury was most often hepatocellular or mixed. The frequency was between 0.1 and 1 % in clinical studies and was by trend lower than for comparators (mostly enoxaparin or warfarin). Comparing the pharmacovigilance reports for the individual NOACs, more hepatic adverse events were reported for rivaroxaban than for dabigatran or apixaban. With the exception of edoxaban, for which only few reports are available, patients with acute liver failure have been reported for every NOAC, but most patients had concomitant drugs or diseases. So far, there are no clear mechanisms explaining the hepatotoxicity of these drugs. We conclude that hepatotoxicity appears to be associated with all NOACs currently on the market. Hepatotoxicity associated with NOACs is idiosyncratic; it appears at therapeutic doses, is rare and the mechanism is not related to the pharmacological action of these drugs. Prescribers should inform patients about possible symptoms of hepatotoxicity and stop these drugs in patients presenting with severe liver injury.