-Kojic acid (KA) has been widely used as a quasi-drug ingredient. Possible promotion activity of KA was suggested on livers of mouse and rat by findings obtained in genotoxicity and carcinogenicity studies performed thus far. Therefore, in order to examine safety as a quasi-drug ingredient, we investigated the presence of initiation activity in rat liver and the photo-genotoxicity and carcinogenicity in mouse skin. In medium-term carcinogenesis test in rats, 2.0% KA was orally given to F344/ DuCrj rats for 4 weeks of the initiation period, followed by the combination of partial hepatectomy and treatment with a hepatocarcinogenesis promoter, phenobarbital (PB). As a result, glutathione S-transferase placental form (GST-P) positive foci of 0.2 mm or more in diameter in the KA group, which is usually used in determination of pre-cancerous lesions, did not increase significantly in both numbers and areas compared with those of the non-initiated controls. In the concurrent analysis, however, numbers of GST-P-positive foci of two cells or more and 0.1 mm or more in diameter increased slightly, and possible weak initiation activity of KA was equivocal. However, considering the known fact that KA exerts promotion activity in the liver of F344 rats by long-term dietary administration, it was suggested that the observed slight increase of the numbers of GST-P-positive foci in rat liver was the effect of promotion activity of KA rather than the initiation activity. In DNA adducts formation assay in a rat liver, no clear adducts derived from KA were detected in male F344/DuCrj rats administered 0.5% or 2% KA orally, and KA was considered not to form DNA adducts in rat liver. In the in vitro photo-reverse mutation assay with bacteria, KA exerted weak photo-mutagenicity. Furthermore, in chromosome aberration study in Chinese hamster lung cells (CHL/IU cells) with UV irradiation, KA induced chromosome aberration at high-dose (1.4 mg/mL) treatment with UV irradiation, but was negative without UV irradiation. However, in the in vivo photo-micronucleus study in mouse, in which 1.0 or 3.0% KA containing cream was applied twice to the back of the animals with a 24-hr interval, KA did not induce micronuclei in mouse epidermal cells. Based on these results, it is considered that the risk of KA to exert photo-carcinogenicity is quite low in the skin. In skin carcinogenesis bioassay for initiation-promotion potential, 3.0% KA cream formulation was applied to the back of the mouse for 1 week (once a day, total 7 times) and for 19 weeks (5 times a week, total 95 times) during the initiation and the promotion stages, respectively. No skin nodules were observed in any animal skins formed due to KA treatment given in either stage. Therefore, KA is consid- ered not to possess initiation nor promotion activity of skin carcinogenesis. Furthermore, from the above findings, it is suggested that KA is virtually safe as a quasi-drug ingredient.