The paper begins with a brief historical account of the production of cancers through exposure to chemicals, first in humans, then in animals, followed by the identification of chemically defined agents. Such agents have progressively been discovered in the most diverse chemical families. The present classification of chemical carcinogens includes those demonstrated as carcinogenic for humans by adequate epidemiological studies and those considered, on more or less limited basis, as probably or possibly carcinogenic for humans. The first category is divided into three groups: complex mixtures-defined chemicals-industrial operations. The materials and compounds of the last two categories are called "potential carcinogens". The present international trend is to consider them as dangerous for humans as long as their innocuousness has not been demonstrated. The author stresses that the compounds of the three categories are either manufactured or developed by man, or of natural origin. The latter seem, by far, to cause the highest percentage of cancers in humans. Then he develops, in their general outline, the pluridisciplinary methodological approaches in view to revealing the carcinogenic potentialities of compounds to which man can be exposed in various fields (occupational, food, therapeutic, household,...). This leads him to underline the factors of uncertainty involved by the extrapolation to humans of the results of animal experimentation and the importance of observations on groups of humans. He also stresses the multifactors which intervene in the various steps of evolution towards malignancy (initiation, promotion, influence of synergistic or antagonistic factors-role of "free radicals" the type of which is the superoxyde anion radical O2-.). He emphasizes the difficulties of interpretation resulting from the multifactorial nature of the processes of malignancy, as well as the interest of revealing the intimate mechanisms of activity (genotoxic mechanisms with the transformation of certain genes: protooncogenes to oncogenes, epigenetic mechanisms); this demonstration becomes more and more performant thanks to the spectacular progress of basis sciences to which cancer evaluation methodology appeals more and more. Evaluation of carcinogenic risks which, together with early diagnosis, constitutes a fundamental basis of cancer prevention is thus considerably improved. The author is thus led to discuss the possibility of setting threshold values, at least for some types of carcinogenic chemicals. In the end, he insists upon the necessity of an active international cooperation for carcinogenicity studies, either experimental or epidemiological.