Biophysical studies of different ionizing radiations and their differences in biological effect can provide useful information and constraints on the nature of the initial biologically relevant damage and hence the subsequent biochemistry and repair processes. It is clear that the nature of the predominant critical component produced by densely ionizing (high-LET) radiations is qualitatively, as well as quantitatively, different from that which predominates for low-LET radiations. Comparisons of radiation track structure with observed biological effects of the radiations allow hypotheses to be developed as to the nature of these different types of damage. That associated with low-LET radiations seems consistent with what is known about DNA double-strand breaks (dsb). It is produced predominantly by a localized cluster of ionizations within a single electron 'track end' either by direct action on the DNA or in conjunction with closely-associated molecules. The characteristic high-LET damage is somewhat larger in number of ionizations and spatial extent and therefore presumably also in molecular complexity. It is suggested that the total spectrum of initial damage be categorized into four classes; in addition to the above two this would include on the one extreme sparse isolated ionizations, which may lead to very simple products that are of limited biological relevance, and on the other extreme very large and relatively rare events which are uniquely achievable by some high-LET radiations, such as alpha-particles, but not at all by low-LET radiations. These biophysical considerations pose a challenge to radiation chemistry studies to consider the chemical consequences of highly localized clusters of initial ionizations and excitations in or very near to DNA, and to biochemistry to consider classes of damage involving DNA (and perhaps associated molecules) of greater complexity than the simplest dsb.