Using a nose-only inhalation system, male Sprague-Dawley rats were exposed 4.2 h d(-1), 5 days per week for 65 weeks to one of two concentrations of natural uranium ore dust aerosol (44% U, 50 mg m(-3) and 19 mg m(-3)) without significant radon content. After inhalation exposure ceased, the rats were allowed to live for their natural lifetime. Lung uranium burdens, measured at the time of death of each animal, declined exponentially after dust inhalation ceased, and the rate of decline was independent of the initial lung burden. Lymph node specific burdens ranged from 1 to 60 fold greater than the specific lung burden in the same animal. No lymph node tumors were observed. The frequency of primary malignant lung tumors was 0.016, 0.175 and 0.328 and primary non-malignant lung tumors 0.016, 0.135 and 0.131 in the control, low and high aerosol exposed groups, respectively. There was no difference in tumor latency between the groups. Absorbed dose to the lung was calculated for each animal in the study. The average doses for all the animals exposed to the low and high dust aerosol concentrations were 0.87 Gy and 1.64 Gy respectively, resulting in an average risk of malignant lung tumors of about 0.20 tumors per animal per Gy in both groups. The frequency of primary lung tumors was also calculated as a function of dose increment for both exposed groups individually and combined. The data indicate that, in spite of the above result, lung tumor frequency was not directly proportional to dose. However, when malignant lung tumor frequency was calculated as a function of dose rate (as measured by the lung burden at the end of dust inhalation) a direct linear relationship was seen (p < 0.01) suggesting dose rate may be a more important determinant of lung cancer risk than dose. Conversely, non-malignant lung tumors were significantly correlated with low lung burdens (p = 0.01). We conclude that chronic inhalation of natural uranium ore dust alone in rats creates a risk of primary malignant and non-malignant lung tumor formation and that malignant tumor risk was not directly proportional to dose, but was directly proportional to dose rate.