The nature of at least three stages in carcinogenesis by chemical agents justifies the existence of a threshold dose: (1) a cytoplasmic stage during which most carcinogenic molecules are eliminated, (2) a nuclear stage during which certain DNA lesions are repaired and therefore cannot help to bring about mutations, and (3) an extracellular stage during which mutations are controlled for a long time by positive or negative epigenetic factors. All these findings are very difficult to collate. Since threshold doses cannot be detected by direct experimentation, the knowledge that a threshold exists is of little practical use and does not much alter the data permitting extrapolation of the results obtained with high doses of carcinogens. At the theoretical level only, this conclusion renders obsolete the opinion that all carcinogenic agents should be completely eliminated from our environment, and also the view that DNA needs only one hit from such an agent for a carcinogenic effect to be produced. The above conclusion also provides a more rational basis for the concept of a 'virtually safe dose' (VSD) or 'allowable daily intake' (ADI).