To clarify the in vivo genotoxic potential of kojic acid (KA), formation of DNA adducts and 8-hydroxy-deoxyguanosine (8-OHdG) in the thyroids of male rats subjected to dietary administration of 2% KA for 2 weeks were assessed by 32P-postlabeling analysis and with a high-performance liquid chromatography system coupled to an electrochemical detector (ECD), respectively. In addition, to investigate possible tumor initiation activity, male F344 rats were given diet containing 0, 0.02, 0.2 or 2% kojic acid for 8 weeks followed by administration of 0.1% sulfadimethoxine (SDM), a thyroid tumor promoter, in the drinking water for 23 weeks with a subsequent 13-week recovery period (two-stage thyroid tumorigenesis model). Rats given four times by s.c. injection of N-bis(2-hydroxypropyl)nitrosamine (DHPN; 700 mg/kg bw) during the initiation period followed by administration of 0.1% SDM and rats given diet containing 2% KA for the initial 8 weeks or for the entire 31 weeks of the experiment, or basal diet alone were provided as controls. DNA adducts were not formed, and the 8-OHdG level was not increased in the thyroids of rats given 2% KA for 2 weeks. In the two-stage thyroid tumorigenesis model, neither adenomas nor carcinomas were induced in the groups given 0, 0.02, 0.2 or 2% KA followed by 0.1% SDM administration, and incidences and multiplicities of focal follicular cell hyperplasias did not demonstrate any significant intergroup differences at the end of administration and recovery periods. In contrast, incidences and multiplicities of focal follicular cell hyperplasias, adenomas and carcinomas were all significantly increased in the DHPN + 0.1% SDM group. Although the incidences and multiplicities of focal follicular cell hyperplasias in the group given 2% KA for 31 weeks were greater than those in the 2% KA + 0.1% SDM group and an adenoma was observed in a rat at the end of the recovery period, no development of carcinomas was evident at either time point. No thyroid proliferative lesions were induced in the group given 2% KA for the initial 8 weeks only. The results of the present studies indicate that KA has neither in vivo genotoxic potential nor tumor initiation activity in the thyroid, and strongly suggest that the earlier observed thyroid tumorigenic activity of KA is attributable to a non-genotoxic mechanism.