A comprehensive understanding of the radiobiological bases of total body irradiation (TBI) is made difficult by the large number of normal and malignant tissues that must be taken into account. In addition, tissue responses to irradiation are also sensitive to associated treatments, type of graft and a number of patient characteristics. Experimental studies have yielded a large body of data, the clinical relevance of which still requires definite validation through randomized trials. Fractionated TBI schemes are able to reduce late normal tissue toxicity, but the ultimate consequences of the fractional dose reduction do not appear to be equivocal. Thus, leukemia and lymphoma cells are probably more radiobiologically heterogeneous than previously thought, with several cell lines displaying relatively high radioresistance and repair capability patterns. The most primitive host-type hematopoietic stem cells are likely to be at least partly protected by TBI fractionation and may hamper late engraftment. Similarly, but with possibly conflicting consequences on the probability of engraftment, the persistence of a functional marrow stroma may also be fractionation-sensitive, while higher rejection rates have been reported after T-depletion grafts and fractionated TBI. In clinical practice (as for the performance of relevant clinical trials), the influence of these results are rather limited by the heavy logistic constraints created by a sophisticated and time-consuming procedure. Lastly, clinicians are now facing an increasing incidence of second cancers, at least partly induced by irradiation, which jeopardize the long-term prospects of otherwise cured patients.