There is a need for better biomarkers of drug-induced liver injury (DILI) to guide risk assessment and patient management. Over the past 3 years, both animal and clinical studies have provided proof-of-concept data showing that a subset of miRNAs appear to offer unique advantages over the conventional DILI biomarkers, such as enhanced sensitivity and specificity, reduced inter-individual variations, the potential to differentiate lethal and nonlethal liver injury, and the ability to reflect the patterns and even the etiology of liver injury. Notably, many studies have demonstrated that level of miR-122, a liver-enriched miRNA accounting for approximately 70% of total hepatic miRNAs, was increased many fold in the blood when DILI occurred. However, currently available data are predominantly based on animal models and not human samples. Due to the lack of a standard quantification method for miRNAs and confirmatory studies using a comprehensive list of drugs and patients, the true value of all reported miRNA biomarkers remains to be carefully assessed. An outstanding challenge is to examine if miRNAs are also useful for idiosyncratic DILI, which constitutes the major part of clinical DILI cases but generally cannot be recapitulated in traditional animal models or in clinical trials (the latter due to its relative rarity).