Summary Several studies have revealed an excess of malformations in children with certain malignancies. A few environmental causes have been identified which may damage the foetus and lead to malformation and cancer. However, most of the numerous recognised cancer/malformation syndromes are genetically determined.This report describes a case-control study of 555 newly diagnosed children with cancer and 1,110 matched controls, chosen from general practitioner lists (GP controls) and hospital admissions (H controls). Their parents were interviewed on topics of possible aetiological significance and medical records were checked to confirm reports at interview. The numbers of congenital malformations in the index and GP control children, and the relatives of the index children, the GP and H controls are described.There were more children with malformations among the cases (60/555) than among the GP controls (27/555), P<0.001. The abnormalities in the cases included eight with specific chromosomal/genetic conditions (e.g. Down On the whole genetically determined factors seem to be more important than environmental agents in the aetiology of childhood cancers, although they may interact with each other. Knudson proposed a 'two mutation' or 'two hit' hypothesis for the development of retinoblastoma (Knudson, 1971;Hethcote & Knudson, 1978;Knudson, 1978) which has since been shown by DNA technology to be correct (Cavenee et al., 1983) and may apply to several other childhood cancers too. However, although a large number of specific anomalies has been described in association with certain childhood cancers (Schimpke, 1978;Knudson, 1986;Dodet & Lenoir, 1990) such as sporadic aniridia, genito-urinary anomalies and mental retardation with Wilms' tumour due to chromosome lIp 13 deletions (Riccardi et al., 1978) there have also been reports of an excess of various less specific anomalies in children with Wilms' tumour (Miller et al., 1964), germ cell tumours (Birch, 1980;Birch et al., 1982;Johnston et al., 1986), rhabdomyosarcomas (Ruymann et al., 1988 and liver tumours .The purposes of the investigation described in this report were to quantify the association of congenital malformations with childhood cancers, and, by studying the incidence of malformations in relatives, to determine to what extent these were inherited and to what extent they might represent new mutations. In addition, family pedigrees were examined for evidence of known and hitherto unrecognised cancerassociated syndromes. The parents of 555 children who were newly diagnosed to have cancer and resident in the West Midlands, North West and Yorkshire Health Authority Regions in England were interviewed with a standard questionnaire. Each case child was matched for age and sex with two control children selected from general practitioner lists, designated GP controls (GPC), and hospital admissions, designated hospital '