SUMMARYThe inter-regional epidemiological study of childhood cancer analysed data on 234 children diagnosed with leukaemia or lymphoma and 468 controls matched for age and sex. A wide range of potential risk factors was examined, including prenatal exposure to x rays, maternal drug ingestion and smoking, child's medical history, and parental medical conditions and occupation. Calculations were completed for leukaemia or lymphoma and diagnostic subgroups, as defined by laboratory confirmed cell type. In utero exposure to narcotic analgesics was weakly associated with leukaemia or lymphoma but no other antenatal factors gave significant risks. New associations were identified for skin diseases in both parents and congenital abnormalities in the mothers of children with leukaemia. For past medical conditions in the child, viral disease occurring under 6 months of age increased the risk for acute lymphoblastic leukaemia. Fewer children in the leukaemia or lymphoma group had been immunised compared with the control groups. Case children diagnosed over the age of 9 years were more likely than controls to have had four or more previous episodes of illness. Overall, these results indicate that prenatal factors may be less important than postnatal or genetic influences in the development of leukaemia or lymphoma in children.In the United Kingdom data have suggested a rise in the incidence of childhood leukaemia.' 2 The reasons for this increase remain obscure, in parallel with the failure of epidemiological studies to detect any major risk factors for childhood leukaemia or lymphoma. Advances in immunophenotyping of haematologic malignancies now permit the identification of distinct subtypes of diseases once thought to be single entities. Hence aetiological investigation of accurately classified disease phenotypes can be accomplished. As yet few epidemiological studies have incorporated this approach, particularly for childhood reticuloendothelial neoplasms.