Previously, we established the in vivo lung metastasis model of rat HCC induced by two hepatocarcinogens, diethylnitrosamine (DEN) and N-nitrosomorpholine (NMOR) at a dose of 120 ppm. This model allows us to investigate modifying factors leading to the inhibition of metastasis formation. However, low survival rates made the evaluation of metastasis formation difficult. The current experiments were conducted to modify the experimental protocol to improve survival and to establish a better animal metastasis model. Lower doses of NMOR (80 or 40 ppm in drinking water) were given to F344 rats for 14 weeks after DEN treatment. Survival rates in the 80 ppm group and in the 40 ppm group were 57% and 81%, respectively and these values were significantly higher than that in 120 ppm. Incidences of lung metastasis in the 40 ppm group steadily increased up to 67% by week 36 while that in the 80 ppm increased sharply up to 86% by week 24. Severity of lung metastases in the 40 ppm group at week 36 was mild compared with the 80 ppm group at week 24. In the second experiment, in order to characterize HCC development and lung metastasis in the 40 ppm group, rats given DEN and then followed with 40 ppm NMOR were killed sequentially. Development of HCC was observed at week 14 and reached 100% incidence at week 20. First lung metastatic lesions were evident at week 22, and incidence of lung metastasis reached 100%. Tumor cells were identified in the blood at week 20 by RT-PCR. The current study revealed that 40 ppm NMOR for 14 weeks after DEN treatment developed HCC without lung metastases at week 22, then HCC with a frequent lung metastasis at week 40. Thus, it can be said that this system is a more appropriate model for elucidation of mechanisms of metastasis and also for analysis of factors to inhibit natural metastasis.