We and others have postulated that modifications to epidemiological methods and tools could improve understanding of childhood cancer aetiology. We describe features of paediatric cancer that influence study design and data collection strategies, and examine determinants of the reproducibility and accuracy of mothers' responses to questionnaires, which are the primary source of risk factor information. Two focus group sessions with mothers of children with acute lymphoblastic leukaemia (ALL) and two with mothers of children with brain tumours were conducted to explore: the optimal time of day, method of administration, and location for the interview; the availability of alternative data sources; the interval between paediatric cancer diagnosis and epidemiological interview; and other features which may affect maternal interview responses. Mothers of children with both types of cancer preferred being interviewed at home during the evening and were willing to provide complete access to their offsprings' medical records. Emotional, socio-economic and perhaps cultural differences between the groups of mothers were exemplified by the willingness of mothers of children with ALL to participate in epidemiological interviews earlier in their child's treatment course and to provide greater access to maternal reproductive history records compared with mothers of children with brain tumours. Parental concerns about the difficult disease and treatment course of children with brain tumours were a key element in their decision to defer participating in epidemiological interviews for many months after their child's cancer diagnosis. We conclude that the focus group approach can contribute to a broad strategy for improving questionnaires and methods for conducting paediatric cancer epidemiological research.