Information on 16,193 infants delivered in Great Britain in one week of April, 1970 was collected by midwives at the birth and during the first 7 days of life. Using multiple sources, 33 children developing cancer by 1980 were identified from this cohort, giving an incidence of 2.04 per 1,000 total births by the age of 10. Comparisons of these 33 children were made with 99 controls, three for each index case, matched on maternal age, parity and social class. Statistically significant associations were initially found with maternal X-rays and smoking during pregnancy, and the use of analgesics such as pethidine during labour, confirming the findings of retrospective case-control studies. Unexpected statistically significant associations were found with delivery of the child outside term, and drug administration in the first week of life. The latter was found in the absence of an association with neonatal abnormalities in the child and relates mostly to the administration of prophylactic drugs such as vitamin K. Logistic regression involving the whole cohort showed independent statistical associations with maternal smoking (OR 2.5), and drugs to the infant (OR 2.6). After adjusting for these factors no other statistically significant associations were found.