[en] Various hypotheses for radiation local tumor control probability (ltcp) were modeled, and assessed against local tumor control (LTC) and clonogenic assay (CA) data. For head-and-neck tumors receiving low-LET external-beam irradiation, the best model was a Poisson ltcp accounting for cell repopulation, hypoxia, and tumor volume dependence of radiosensitivity (α). This confirmed that hypoxia is limiting LTC of these tumors, with the magnitude depending upon tumor volume. However, LTC of cervical carcinoma receiving external-beam irradiation and brachytherapy was well described by a model not accounting for hypoxia. Furthermore, when the survival fraction at 2 Gy (SF₂) and colony forming efficiency (CFE) measured for individual patients were incorporated into this model, very good correlation with LTC was seen (p=0.0004). After multivariate analysis, this model was the best independent prognostic factor for LTC and patient survival. Furthermore, no difference in prediction was seen when a model based on birth-and-death stochastic theory was used. Two forms of hypoxia are known to be present in tumors: diffusion-limited, chronic hypoxia (CH), and acute, transient hypoxia (TH). A modeling study on WiDr multicellular spheroids showed that the CH effect on LTC is significantly lower than expected from CA. This could arise from energy charge depletion accompanying CH, reducing the number of proliferating clonogenic cells that can repair radiation damage, and thus mitigating the radioresistance of CH cells. This suggests that TH, rather than CH, may be the limiting factor for in vivo LTC. Finally, by computing ltcp using Monte Carlo calculated dose distributions, it was shown that Monte Carlo statistical noise can cause an underestimation of ltcp, with the magnitude depending upon the model hypotheses