Background: The safety of herbal products is an important issue in Korea. It is more complicated because of the concomitant use of conventional western drugs and herbal medicine, including prescribed medicine, folk herbs and food supplements. Although both western and traditional Korean medical doctors have studied on the safety of HMP, their results did not show relevant or consistent conclusions because of the poor quality of studies, including bias. Objectives: The aim of this study was to review the studies in Korea related to drug induced liver injury (DILI) systematically. We tried to estimate the proportion of causative materials of DILI and evaluate the clinical difference of DILIs from different materials. Methods: Systematic searches were performed on MedRIC and MEDLIS in Korea. The extraction of data as well as selective screening of the studies was carried out independently by two of the authors. There were no restrictions on the types of publication, including grey literature. Results: Twenty-one articles were included. Of them only 7 adapted prospective design and only 6 were published in peer-reviewed journals. The proportion of conventional drugs associated with hepatotoxic injuries in all DILI ranged from 15.8% to 83.3% and that of herbal medicine was from 28.9% to 44.7%. However, the criteria for herbal medicine were not clearly defined and concomitant medications were not fully investigated in most studies. There were limited objective data in the clinical differences of liver injuries from different materials and their results were conflicting. Conclusions: The causality assessments regarding DILI of herbal medicine were not performed properly and causative materials were misclassified in most of the studies published in Korea. These make the safety issue still ambiguous because of the limitations and lack of objectivity of the studies. More rigorous studies are required for clearly addressing these conflicting issues with cooperative investigation between traditional Korean and western medicine.