RÉSUMÉCes deux dernières années, les principaux comités internationaux travaillant sur les effets des faibles doses de rayonnement ont publié une nouvelle série d'indicateurs du risque en fonction des doses reçues. Ces travaux intègrent les évolu-tions des données épidémiologiques, notamment depuis la révision de la dosimetric des personnes exposées à Hiroshima et Nagasaki. ABSTRACT During the last two years, international committees dealing with the effects of ionizing radiation have published a new series of risk estimates. New epidemiological data, in particular In the case of A-bomb survivors whose doses have been reassessed, have been used. Unfortunately, there has been a large range of alternatives, not only in dealing with epidemiological or biological modelling, but also in the actual computation of a life-long risk index which is derived from the epidemiological "primary coefficients". As a consequence, published figures are never truly comparable. The identification of all computational alternatives is made here, and the quantification of their consequences is attempted. The purpose is to achieve the control of demographic assumptions (e.g. reference population) and of various conventional assumptions, in order to measure the impact of epidemiological and biological hypotheses, which are felt to be more fundamental. The analysis shows that such impacts are more important than suggested by published tables. The effects of various alternatives obviously compensate one another. Further discussions on the modelling of the biological effects of radiation would greatly benefit from the development of a "standar" for life-long risk computations.