Four thousand eighty inbred rats were maintained from weaning on various different concentrations of N-nitrosodiethylamine (NDEA) or N-nitrosodimethylamine (NDMA). The principal aim was to characterize the dose-response relationship for the effects of these agents on esophageal cancer (NDEA) or on various types of liver cancer (NDEA and NDMA), although NDEA also caused a few tumors of the nasopharynx and NDMA also caused a few tumors of the lung. The numbers of tumors of mesenchymal and Kupffer cells in the liver were too few to allow easy characterization of the dose-response relationships, and although NDMA induced large numbers of bile duct neoplasms, NDEA did not. Thus, the four principal dose-response relationships studied were of NDEA on esophageal or liver cells and of NDMA on bile duct or liver cells. At doses sufficiently high for the median time to death from the disease of interest to be estimated, relationships were observed of the general form (Dose rate) x (median)n = constant where n was about 2.3 for the first three relationships and about 1 for the last one (NDMA on liver cell tumors). By contrast, at doses sufficiently low for longevity to be nearly normal (median survival about 2.5 years), there remained no material dependence on the dose rate of the age distribution of the induced neoplasms. At these low dose rates, the number of liver (but not of esophageal) neoplasms induced by treatment was simply proportional to the dose rate. This finding is not surprising, since the background incidence of liver (but not of esophageal) neoplasms was appreciable. The linear relationship observed at low dose rates (below 1 ppm) suggests that under these experimental conditions, among rats allowed to liver their natural life span, a dose of 1 ppm of NDEA or NDMA in the drinking water will cause about 25% to develop a liver neoplasm, a dose of 0.1 ppm will cause about 2.5% to do so, and a dose of 0.01 ppm will cause about 0.25% to do so, etc., with no indication of any "threshold." (At these low dose rates, the incidence of liver neoplasms appears likely to exceed greatly that of esophageal neoplasms.) In addition, even quite low dose rates of the test agents caused a variety of nonneoplastic liver abnormalities (e.g., hyperplastic nodules, or shrinkage of hepatocytes) at a frequency roughly proportional to the dose rate.