Targeted Radionuclide Therapy (TRT) consists in killing tumour targets by using radiolabeled vectors (radiopharmaceuticals) that selectively bind to tumour cells. In a context of TRT optimization, a better determination of energy deposition within biologic material is a prerequisite to the definition of the absorbed dose-effect relationship and the improvement of future cancer treatment. This requires being able to quantitatively assess activity distribution (with the most appropriate molecular imaging technique) and perform radiation transport at the scale at which biologically relevant phenomena occur. The methodologies that should be applied and the problematic to be faced strictly depend on the scale (cell, tissue, body) of the application considered, and on the type of radiation involved (photons, electrons, alpha). This research work consisted in developing dedicated dosimetric techniques (single-scale dosimetry) capable of taking into account the peculiarity of different experimental scenarios (cellular, pre-clinical, clinical TRT). All methods developed were tested in the framework of actual research applications and experiments: • The development and validation of a 3D cellular model allowed a better understanding of the radiative and non-radiative processes associated to cellular death in the case of clonogenic survival experiments involving beta emitters. • A Monte Carlo based application for the calculation of the absorbed dose distributions in ex-vivo mouse tumours helped in assessing the absorbed dose- effect relationship for three different antibodies labelled with an alpha emitter (212Pb). • An adaptive resolution approach to clinical dosimetry (multi-scale dosimetry) is also proposed in order to increase the accuracy of absorbed dose delivery in small radiosensitive organs.