It has long been known that certain estrogens and testosterone may increase, or sometime decrease, the incidence of neoplasmas in laboratory animals. They probably act by switching-on inappropriate genetic information or switching-off appropriate genetic information contained in nucleic acids. For instance, they may switch-on in adult animals information which is relevant only to a certain stage of embryogenesis or they may facilitate the expression of oncogenic viruses which would otherwise lay harmlessly dormant. The situation is rendered complex because an effect on one endocrine gland leads to effects on others so that factors which favour tumour development may result indirectly from administration of an anabolic or other hormonal agent. Two kinds of neoplasm are now known to be associated with human exposure to anabolic agents: vaginal adenocarcinoma and liver-cell tumours. Tumours of both kinds are among the spectrum of neoplasms that has been seen in laboratory animals exposed to agents of the same kind. In both animals and man there is evidence that tumours arising in response to anabolic agents are sometimes, initially at least, hormone-dependent. The evidence that 17beta-estradiol, diethylstilbestrol, chlormadinone, and testosterone are carcinogenic for laboratory animals is briefly reviewed and the reader's attention is directed towards the 1974 IARC Monograph on the evaluation of sex hormones for carcinogenic risk to man where the same evidence is reviewed more extensively. The significance for man of the results of studies on laboratory animals is discussed with special reference to the use of anabolic agents in meat production. Non-residue uses are to be preferred, as are naturally-occurring agents as opposed to compounds which do not occur in nature. More information is needed concerning the possible effects of prolonged exposure to very low doses of anabolic agents.