The United Kingdom Childhood Cancer Study was designed to examine the potential aetiological role of a range of perinatal and reproductive factors. Our use of clinical records permitted a more exact characterisation of reproductive events than is possible in investigations that rely on self-reporting; and the increased specificity with which antecedent events were measured produced more precise risk estimates, albeit ones based on progressively smaller numbers. Information on the conduct of this component of the study and results for 1485 children with haematological malignancies and 4864 controls are presented. The 'find' rate for obstetric records was high at 86% for cases, with 81% having information on both matched controls. Associations were seen for severe hyperemesis (Odds Ratio=3.6, 95%Confidence Interval=1.3-10.1, for all leukaemias), polyhydramnios (OR=4.0, 95%CI=1.5-10.3, for acute myeloid leukaemia (AML)), anaemia (haemoglobin <10 g, OR=2.6, 95%CI=1.7-4.1, for AML), and pre-eclampsia (OR=1.7, 95%CI=1.1-2.7, for non-Hodgkin's lymphoma). Babies who developed leukaemia were heavier at birth (>4000 g, OR=1.2, 95%CI=1.0-1.4), as were their older siblings (>4000 g, OR=1.4, 95%1.0-1.9). Mothers' whose children developed common B-cell precursor acute lymphoblastic leukaemia (ALL) were more likely to have had a previous molar pregnancy (OR=5.2, 95%CI=1.9-14.7). Gender-specific analysis revealed that findings often differed markedly for boys and girls; and, in common with other reports, strong associations with Down's syndrome were seen for both ALL and AML.