The antitumor effects of melatonin on diethylnitrosamine (DEN)-initiated and/or phenobarbital (PB)-promoted hepatocarcinogenesis were investigated in male F344 rats. Five-week-old male F344 rats were divided into eight groups. Rats in groups 1-5 were given DEN (100 mg/kg body weight, i.p.) once a week for 3 wk, whereas those in groups 6-8 received vehicle treatment. Groups 1-3 and 7 were given 500 ppm PB in drinking water for 20 wk after DEN or vehicle treatment. Group 2 was given 400 ppm melatonin-containing diet during the initiation phase. Groups 3 and 5 were fed melatonin-containing diet for 20 wk, starting 1 wk after the last dosing of DEN. Group 6 was given melatonin-containing diet alone throughout the experiment (24 wk). Group 8 was treated with vehicle alone. Liver neoplasms were recognized only in DEN-treated groups. The incidences and multiplicities of hepatocellular adenoma and hepatocellular carcinoma (HCC) in group 3 were significantly smaller when compared with group 1 (P < 0.001 or P < 0.002). The average and unit areas of glutathione S-transferase placental form (GST-P)-positive foci of groups 2 and 3 were significantly smaller than those of group 1 (P < 0.001 or P < 0.01). The density and average area of these preneoplastic lesions of group 5 were also smaller than those of group 4 (P < 0.001 or P < 0.005). In addition, the ornithine decarboxylase activity in nonneoplastic liver tissue was reduced by melatonin treatment in both the initiation and postinitiation phases. These results suggest that melatonin has an antitumor-promoting ability in DEN-initiated and PB-promoted hepatocarcinogenesis in rats.