The carcinogenic potential of anticancer drugs is discussed in the light of selected basic principles of chemical carcinogenesis. Anticancer drugs which act by alkylation and/or by binding tightly to DNA frequently cause cancer in experimental animals and may be carcinogenic in man. In addition, certain anticancer drugs act as cocarcinogens in experimental systems and augment the tumorigenicity of chemical carcinogens. Host determinants are important in chemical carcinogenesis. Many chemical carcinogens and anticancer drugs require metabolic activation by microsomal enzymes. Studies in twins have shown interindividual variation of drug metabolism in man is greater than intraindividual variation caused by exogenous factors. Therefore, certain individuals may be unusually susceptible to the carcinogenicity of anticancer drugs on a pharmacogenetic basis. Age is also a host determinant. At a given total dose level, age at first exposure to chemical carcinogens has been shown to be an important risk factor in experimental studies and in some epidemiologic investigations in man. Therefore, children may be especially susceptible to the carcinogenicity of anticancer drugs. These treated children have the potential of a normal lifespan; the latency period between initial exposure to a carcinogen and clinical evidence of cancer in man is long, usually 2-5 decades. The problems involved in extrapolating data of carcinogenicity in experimental animals to man are discussed. A single drug may have multiple consequences in experimental studies; for example, actinomycin D can act as an anticancer drug, an anticarcinogen, and a carcinogen. These uncertainities and the clinical results concerning second neoplasms following cancer therapy in both children and adults clearly indicate the need to follow carefully long-term survivors who have received cancer therapy.