Microdosimetric aspects of tritium radiotoxicity are discussed. The level of coincidence of tritium locations and of radiation-sensitive sites does not appear to be decisive for relative effectiveness of tritium exposures. The actual target structure and dimensions seem to be of more importance. Particularly efficient germ cell destruction by chronic exposures to tritiated water is unlikely to he explainahle on the basis of classical principles of microdosimetry. The phenomenological approach of investigating biological response functions may be a useful alternative for the interpretation of unexpected experimental results, especially when combined with bidimensional microdosimetry. The latter allows for clustering of nanometer sites within a target of cellular dimensions.