Today's radiation safety norms are based on the linear no-threshold theory (LNT): extrapolation of the dose-response relationships down to the minimal doses, where such relationships are unproven and can be inverse due to hormesis. The most promising way to obtaining reliable data on the dose-effect relationships for low radiation doses would be large-scale animal experiments. Outstanding published data on carcinogenic effects of the doses e.g. below 100 mSv should be verified by experiments. Arguments against applicability of the LNT to the doses comparable to those from the natural radiation background are discussed. Furthermore it is stressed that medical consequences of the Chernobyl accident have been overestimated; and this theme has been exploited to strangle development of atomic energy and to elevate prices for fossil fuels. Worldwide introduction of nuclear energy will be possible only after a concentration of authority within a powerful international executive. It would enable the construction of nuclear reactors in optimally suitable places, considering all sociopolitical, geographical, and geological conditions, which would contribute to the prevention of accidents like in Japan in 2011. A concluding point is that radiation safety norms are exceedingly restrictive and should be revised to become more realistic and workable. Elevation of the limits must be accompanied by measures guaranteeing their strict observance. It is also concluded that there are no evidence-based contraindications to fivefold elevation of the total equivalent effective doses to individual members of the public (up to 5 mSv/year), and doubling of the limits for professional exposures. © The Author(s) 2012.