From December 2004 to July 2005, three reports on the effects of low doses of ionising radiation were released: ICRP (2004), the joint report of the French Academies of Science and Medicine (Tubiana et al 2005), and a report from the American Academy of Sciences (BEIR VII 2005). These reports quote the same recent articles on the biological effects of low doses, yet their conclusions diverge. The French report concludes that recent biological data show that the efficacy of defense mechanisms is modulated by dose and dose rate and that linear no threshold (LNT) is no longer plausible. The ICRP and the BEIR VII reports recognise that there are biologic arguments against LNT but feel that there are not sufficient biological proofs against it to change risk assessment methodology and subsequent regulatory policy based on LNT. They point out the remaining uncertainties and the lack of mechanistic explanations of phenomena such as low dose hyperlethality or the adaptive response. In this context, a critical analysis of the available data is necessary. The epidemiological data and the experimental data challenge the validity of the LNT hypothesis for assessing the carcinogenic effect of low doses, but do not allow its exclusion. Therefore, the main criteria for selecting the most reliable dose-effect relationship from a scientific point of view should be based on biological data. Their analysis should help one to understand the current controversy.