Solid tumors can be ablated with minimal invasiveness using radio-frequency (RF) heating under guidance of magnetic resonance imaging (MRI). To analyze the thermal damage in tissue produced under different input power and RF probe conditions, we developed a bio-heat model of the three dimensional (3D) temperature distribution dynamics in tissue. In this model, the tissue temperature was increased by RF heating, decreased by perfusion, and distributed by heat conduction. Furthermore, we associated a probability model that relates perfusion to thermal damage locally. Thermal damage (and blood coagulation) was determined by the temperature history at each position in the tissue. During the ablation process, the local perfusion begins to decrease when the temperature rises above 43°C. Our model of thermal damage includes the probability of cell death and recovery from damage. The rate coefficients of this model depend on the local temperature history. The aim of this model was to predict when tumor cells in a local region are truly dead rather than simply damaged. This tissue-damage model provides a fundamental basis for quantifying changes in the perfusion coefficient of the bio-heat equation. We compare model predictions with MR lesion images repeatedly over time when ablation ceases as well as with histological tissue studies.