Epidemiological evidence for prenatal carcinogenesis includes associations between cancer in young people and intrauterine exposure to X-rays, drugs and hormones and prezygotic events such as specific chromosomal aberrations associated with specific cancers. Recent findings suggest that the hormonal environment during early gestation can result not only in the development of clear-cell adenocarcinoma of the vagina but also in the development of germ cell tumours of the testes and ovary. Hormone-related risk factors for testicular germ cell neoplasms include maternal use of exogenous oestrogens during early gestation and, possibly, maternal nausea, maternal obesity and race as well. Ovarian germ cell tumours also appear to be related to maternal use of hormones and obesity. Several epidemiological studies of cancer in young people have been directed towards suggested associations with parental occupational exposures, parental cigarette smoking and household exposures to electric and magnetic fields (EMF). The findings of the many studies of parental occupational exposures are inconsistent and are often nonspecific with respect to the type of childhood cancer and the job exposure implicated. Parental cigarette smoking has been associated in some studies with an increased risk for cancer among children and young adults, and in other studies with an increased risk among mature adults, but the findings are not consistent across studies. Three studies of all types of childhood cancer found risk to be related to household exposures to EMF; in all three, the risk for central nervous system tumours was increased, and in two of the three leukaemia risk as well. A fourth study showed no association between childhood leukaemia and EMF. A hypothesis is proposed which suggests that prenatal and early childhood exposure to N-nitroso compounds (NOC) may be related to the development of primary tumours of the brain in children. Experimentalists have shown that various NOC are potent nervous system carcinogens, particularly when animals are exposed transplacentally. This experimental model and findings from a Los Angeles case-control study (209 pairs) of brain tumours in young people led to the proposed epidemiological hypothesis. Although this and other epidemiological studies of NOC have major limitations, findings from epidemiological studies of congenital defects and of other childhood cancers lend the hypothesis some support. A large international collaborative case-control study of childhood brain tumours was begun recently. This study has a major advantage over most case-control studies in adults because the exposure period of greatest interest (gestation) is clearly defined.