Object: Hadrontherapy is becoming an increasingly attractive modality for cancer treatment due to the favourable depth-dose profile of ions and high relative biological effectiveness (RBE) in the tumour region. Since RBE depends on multiple parameters related both to the irradiation beam and the cell properties, biophysical models are essential to comply with the demands of a clinical environment. NanOx addresses some of the flaws in the models currently implemented in the treatment planning systems, and presents many innovative features. Method: The model takes into account the fully stochastic nature of ionizing radiation by considering dose fluctuations both at nanometric and micrometric scales, and introduces the concept of chemical dose. The latter represents the induction of cell death by “non-local” events as the accumulation of cellular oxidative stress or sub-lethal lesions induced by radical species. Such “non-local” events are complementary to the “local” events, which take place at a very localized scale and are considered as lethal since can singly cause cell death. Results: NanOx predictions for V79, CHO and HSG cell lines irradiated by photons, protons and carbon ions are in good agreement with the experimental data. The model is able to describe the effectiveness of ions, including the overkill effect at high LET values. Moreover, the typical shoulder in cell survival curves is reproduced owing to the introduction of the chemical dose which varies with LET. Conclusion: The promising results obtained with NanOx stress its potential in the context of hadrontherapy, and may lead in the future to apply it to neutron beam therapy or photoactivation of nanoparticles. Despite a rigorous mathematical approach, its implementation remains simple and compatible with the constraints of clinical application. The model relies in fact on the fit of five parameters, and its pragmatic architecture facilitates improvements and optimizations.