The 1984 Windscale study raised concern about a possible association between living in the vicinity of nuclear power plants and childhood cancer. No such effect for all cancers was seen in ecological studies in Germany (1980-1995). Results from exploratory analyses led to a new study. Pre-selected areas around all 16 major nuclear power plants in Germany formed the study area. The design is a matched case-control study; cases are all cancers under five years diagnosed in 1980-2003: 1592 cases, and 4735 controls. Inverse distance of place of residence to the nearest nuclear power plant at the time of diagnosis was used as the independent variable in a conditional logistic regression model. Results show an increased risk for childhood cancer under five years when living near nuclear power plants in Germany. The inner 5-km zone shows an increased risk (odds ratio 1.47; lower one-sided 95% confidence limit 1.16). The effect was largely restricted to leukaemia. The results are compatible with the corresponding subgroups in the previous German ecological studies, with which this study shares most of the cases. They contrast with the lack of an effect observed or expected from other studies due to low doses from routine nuclear power plant operation.