Many predictive factors of tumor radiosensitivity have been described. Number of clonogenic cells, proliferation rate, hypoxia and intrinsic radiosensitivity are usually considered as the main parameters of tumor control. Intrinsic radiosensitivity is correlated in a first approach to the ability of the cell to detect and repair DNA damages, and so integrity of the different pathways involved in this function: PARP-1, XRCC1, ATM, p53, MRN complex or BRCA1... Genetic polymorphisms of some of these genes, found in normal lymphocytes, have been correlated to late toxicity of normal tissues. But, in tumors, because of the difficulty to obtain samplings and heterogeneity, accurate molecular analysis is not possible in many cases, and no valuable test of radiosensitivity exist at this moment. For example, TP53 gene has been evaluated in many studies and results regarding its potential as a predictive factor of tumor sensitivity are conflicting. Surviving fraction at 2Gy (SF2) allowed a global evaluation of sensitivity, but the obtention of this parameter often takes a long time and failed in 20 to 40%. Evaluation of double-strand break repair capacity by immunochemistry quantification of phosphorylated forms of ATM, H2AX or MRE11 is an interesting topic. However, discovery of tumor stem cells in a number of epithelial tumors could revolutionize the understanding of radiosensitivity. Combination of genomic and functional techniques are probably essential to better predict this parameter.