NEED FOR EPIDEMIOLOGICAL OBSERVATIONS That there should still be uncertainty about the number of occupational carcinogens in 1984, 66 years after Yamagiwa and Ichikawa1 found a method for demonstrating carcinogenicity in laboratory animals is, perhaps, surprising; but the laboratory methods for detecting carcinogens are still imperfect tools. The results that can be obtained in vitro or in vivo correlate well with human experience in the sense that one or more of the tests is nearly always positive when an agent has been found to cause cancer in Man. In such circumstances, however, efforts are made to demonstrate carcinogenicity that are far more intense than those normally made in the testing of agents for which no human evidence is available, and no practicable battery of routine tests would have detected all the agents now known to cause human cancer. It must, therefore, be assumed that substances carcinogenic to Man will continue to be missed on occasions, quite apart from