The aetiology of childhood leukaemia remains generally unknown, although exposure to moderate and high levels of ionizing radiation, such as those experienced during the atomic bombings of Japan or from radiotherapy, is an established cause. Risk models based primarily on studies of the Japanese atomic bomb survivors imply that low-level exposure to ionizing radiation, including ubiquitous natural background radiation, also raises the risk of childhood leukaemia. Using two sets of recently published leukaemia risk models and estimates of natural background radiation red-bone-marrow doses received by children, about 20% of the cases of childhood leukaemia in Great Britain are predicted to be attributable to this source. However, for one of these sets of risk models this attributable fraction is materially dependent on how the radiation-induced risk is assumed to be transferred between the Japanese atomic bomb survivors and Western children. Over a range of annual doses representing the range (0.5-2.5 mSv/year) experienced by most populations, the attributable proportion for the preferred risk-transfer model varies between 8 and 30%, with small deviations from a linear relationship that are largely due to the saturation of the model, although again this range of attributable fractions depends on the assumed transfer of risk between populations.