Drug-induced liver injury (DILI) is one of the serious and frequent drug-related adverse events and is classified into intrinsic and idiosyncratic types. Almost all of DILI caused in humans is known to be idiosyncratic type. The estimation of the potential risk for a drug candidate to induce idiosyncratic DILI is important to facilitate the development of new drugs, however, the estimation is difficult from the results of non-clinical toxicity studies using animals. We have previously reported the in vitro combination assay of mitochondrial function and apoptosis using human primary hepatocytes as a useful model for estimation of the risk of idiosyncratic DILI. In this study, to improve the in vitro assay for estimation of the risk of idiosyncratic DILI, we evaluate the usefulness of HepaRG cells for the mitochondrial function assay. We measured the oxygen consumption rate (OCR) as an endpoint of mitochondrial function in HepaRG cells treated with some compounds causing idiosyncratic DILI (troglitazone, leflunomide, ranitidine and diclofenac) and others known not to cause idiosyncratic DILI (acetaminophen and ethanol) and compared the results in HepaRG cells and with those for human primary hepatocytes as previously reported. As the results, HepaRG cells showed comparable or even higher sensitivity for detecting mitochondrial dysfunction than human primary hepatocytes, in all tested compounds. Taking into account these results and many other useful properties of the cells, HepaRG cells are considered to be much more suitable for this mitochondrial function assay than the human primary hepatocytes.