The concept of intrinsic radiosensitivity is now strongly associated with the linear-quadratic (LQ) model which is currently the best and the most reliable method to fit the first three decades of a survival curve for both human fibroblast and human tumor cell lines. This approach has led to the major conclusions that it is the initial part, and not the distal part, of the survival curve which truly characterizes intrinsic cellular radiosensitivity and there is a correlation between the parameters describing mainly the initial part of the survival curve (alpha, SF2, D) and the clinical radioresponsiveness. More accurate analysis with flow cytometry or a dynamic microscopic image processing scanner (DMIPS) has allowed further study of the survival curve which has shown two sorts of substructure. On one hand, the overall survival curve of exponentially growing cells is described by two or more sets of alpha, beta parameters (heterogeneity in radiosensitivity due to the cell cycle). On the other hand, hypersensitivity at very low doses (< 0.5 Gy) followed by an increase of the radioresistance of the whole population at higher doses has also been observed. This phenomenon is not described by the conventional LQ model and has been interpreted as an induced radioresistance which seems to be negatively correlated with intrinsic radiosensitivity. In clinical radiotherapy, there are two sorts of response of normal tissues: (1) the early and late damage and (2) the carcinogenesis. Concerning the first point, the clinically detectable radiation damage appears at doses usually around 20 Gy (in 2-Gy fractions) with the exception of the hemopoietic and the lymphatic tissues. Therefore, the small doses delivered at the edges or in the penumbrae of treatment fields in routine radiotherapy cannot create detectable damage, despite a potentially much higher effect per unit dose, because the total doses are still very small. However, it may be important to bear in mind the possible extra effect of low doses outside the target volume if regions in the vicinity are subsequently retreated. Concerning clinical radiation-induced carcinogenesis, three studies described a higher relative risk associated with small doses per fraction or very low dose rate. The results and the interpretation of these studies are discussed.