Studies of natural radiation and childhood cancer must be very large if they are to have sufficient power to detect the very small radiation effects expected. Conventional interview-based case control studies with a sufficiently large number of study subjects would be exceptionally expensive and would also be liable to bias which might lead to confounding that would dominate the results of the study. Record based studies, in which cases and controls are drawn from existing registers, have the potential to be large enough and to avoid the risk of bias. However, they will not have interview-based information on factors such as Socioeconomic Status, nor will they have direct measurements of radiation levels in the homes of study subjects. A large record-based case control study from Great Britain is reviewed here in the context of such case-control studies generally. This study has detected a statistically significant association between indoor gamma ray exposures and the incidence of childhood leukaemia. The risk factor was compatible with that extrapolated from higher dose studies, in particular those of the Atomic Bomb Survivors.