[en] The widespread feeling of 'radiophobia' by the general public has its basis on the ICRP's 'linear no-threshold' hypothesis in dose-response relationship for low dose radiation from the standpoint of radiation protection. Although this common feeling served as a merit for constructing the 'safety culture' of society, it has now become a large obstacle for the development of peaceful uses of nuclear technology as a demerit. Recently many data have been accumulated for the radiation effects of low dose, both epidemiologically and experimentally. Although in general it is very difficult to obtain clear evidence of presence or absence of threshold, it seems to be true that the risk by radiation exposure at low level (the definition of which is below 0.2 Gy) is not so large as that of extrapolation from the high or medium dose range. In fact, many data suggest that some quite different mechanisms are working in low dose from high dose, such as 'adaptive response', and a new concept, 'radiation hormesis', has emerged, that the low level radiation is not only quite harmless but is rather necessary for living cells or beneficial for human health. In this paper, some critical comments on ICRP recommendations are given as a personal view by the author. These include: (1) a question of exact assessment of exposed dose by A-bomb survivors used for the epidemiological data, which are regarded to be the most authentic and important; (2) a brief summary of effects at the natural radiation level, including the high background area data; (3) the importance of dose rate effect, which reflects the living matter's repairability from radiation injury, and (4) the proposal of new paradigm by adopting the reasonable 'de minimis' level (below which there is no harm) both for low dose and at low dose rate. A simple mathematical analysis for representative data of dose rate effect was shown as an appendix