By inspection of the FASSET database on radiation effects on non-human biota, one of the major difficulties in the implementation of ecological risk assessments for radioactive pollutants is found to be the lack of data for chronic low-level exposure. A critical review is provided of a number of extrapolation issues that arise in undertaking an ecological risk assessment: acute versus chronic exposure regime; radiation quality including relative biological effectiveness and radiation weighting factors; biological effects from an individual to a population level, including radiosensitivity and lifestyle variations throughout the life cycle; single radionuclide versus multi-contaminants. The specificities of the environmental situations of interest (mainly chronic low-level exposure regimes) emphasise the importance of reproductive parameters governing the demography of the population within a given ecosystem and, as a consequence, the structure and functioning of that ecosystem. As an operational conclusion to keep in mind for any site-specific risk assessment, the present state-of-the-art on extrapolation issues allows us to grade the magnitude of the uncertainties as follows: one species to another > acute to chronic = external to internal = mixture of stressors > individual to population > ecosystem structure to function.