It is argued that non-genotoxic mechanisms may be more important than genotoxic mechanisms in the genesis of cancer and that the use of tests for genotoxicity to screen chemicals for carcinogenicity in the hope of being able to prevent human cancer is ill-based. The general applicability of the two-stage model for carcinogenesis is also questioned. It is argued that tumour promotion as a second step in a two-stage process is only one of many possible non-genotoxic mechanisms involved in carcinogenesis and not a very important one at that. Many cancers arise against a background of prolonged disturbance of hormonal status leading to hypertrophy and/or hyperplasia of the tissues in which cancers eventually arise. This sequence of events does not fit the two-stage paradigm according to which genetic-damage (initiation) is followed by the stimulation of cell proliferation (promotion). In hormonal carcinogenesis cellular proliferation precedes genetic damage. It is suggested that the genetic damage that occurs as a secondary event in hormonal carcinogenesis may be due, not to exposure to environmental (exogenous) mutagens, but to endogenous mutagens generated during normal metabolic processes in hyperplastic and hypermetabolic cells. Perhaps the rate of damage to DNA in such cells overwhelms the capacity for DNA repair. Cancer is an ageing-related disease. The incidence and severity of both non-neoplastic and neoplastic lesions are greatly enhanced by over-nutrition in laboratory rats. Some of the neoplasms which appear in increased incidence in ad libitum-fed as distinct from slightly (e.g. 80% of ad libitum) diet-restricted rats are of endocrine glands and their appearance is preceded by clear evidence of hormonal disturbance. However, diet restriction protects against malignant tumours of many non-endocrine tissues, e.g. the skin and subcutaneous tissues. It is suggested that diet restriction reduces cancer risk and ageing-associated disease by lowering metabolic rate, reducing the endogenous production of oxygen-free radicals and thereby reducing the rate of DNA damage. These deliberations lead to two recommendations. Firstly much more attention should be paid to the contribution of disturbed physiological status to increased cancer risk especially if the disturbance leads to cellular proliferation, increased cell turnover or increased metabolic rate in particular cells or tissues. In other words there is a great need for carcinogenesis experiments in which there is no deliberate exposure to genotoxic carcinogens. Secondly if we are to have any real hope of relating the results of tests of chemicals for genotoxicity to the results of life-time in vivo carcinogenicity tests, we must as far as possible avoid the confusion caused by premature ageing and enhanced ‘spontaneous’ tumour incidence which arises non-specifically as a consequence of over-nutrition.