The objective of radiation dosimetry both at organ and cellular levels, as applied to radiation protection and radiobiology, is to establish dose-effect relationships that will be helpful for setting appropriate radiation protection standards. Internal Radiation dosimetry plays an important role in nuclear medicine, targeted radio-therapy and radiation protection. In the absence of direct in-vivo measurements of the absorbed doses in human organs, Monte Carlo techniques offer reliable dose estimation in such situations. In the present study we employ Geant4.9.6 simulation tool kit for internal dose estimations. This dissertation is divided into three parts. In the first part benchmarking and validation of Geant4 physics models have been performed. The Geant4 cross section data validation has been performed for various electromagnetic physics models extensively and compared with NIST and ICRU 37 data libraries for a range of energies. It has been found that the relative difference of Geant4 and NIST XCOM data remains within 4.2%. Similarly, percentage differences were up to 1.6% betweenGeant4 and ICRU report 37 data for water. Considering thyroid dosimetry, an experimental procedure has been adopted for benchmarking of Geant4. For regulatory and radiation protection purposes the exposure from radio-iodinated thyroid has also been determined in this work experimentally using patients and phantoms. Comparison of experimentally measured values at 0.5 and 1m distance from neck phantom using ionization chamber, with Geant4 results show a good agreement, with maximum relative differences were up-to 8.4%. The second part consists of development of anthropomorphic phantom for Pakistani population and estimation of dosimetric parameters at organ level. The absorbed fraction values have been estimated for electrons and photon distributed uniformly in spherical, ellipsoidal and cylindrical geometrical models. The energy range adopted in this study covers most of the energies emitted by radio-nuclides currently employed in nuclear medicine procedures or any accidental release of radio-nuclides. Further simulations have been carried out for water, ICRP soft, brain, lung & ICRU Bone tissues as material for these models, considering the elemental composition of each material. ix Thyroid dosimetry for 131I has been performed for various age groups including developing fetus, newborn baby, one, five, ten, fifteen years and adults individuals. The results of S-values (mean absorbed dose rate per unit cumulative activity) calculations are affected by the degree of detail included in the model compared with the original thyroid. Iodine dosimetry has been performed for single and double lobes ellipsoidal model and for anthropomorphic mathematical phantom model in Geant4 simulation. It has also been observed that in the case of 131I β-particles absorbed fraction values increase from 0.88 to 0.97 for developing fetus (10 week to 36 week) which is smaller than ICRP over estimated values. The mathematical anthropomorphic phantom for thyroid employed in Geant4 shows a relative difference 4.3% with ORNL published S-values. An anthropomorphic phantom similar to ORNAL and MIRD stylized phantoms for whole body, has been developed for regional specific (Pakistani) population and has been implemented in Geant4. The Specific Absorbed Fraction values (SAF) has been estimated for both male and female vital organs, considering an energy range of 10 keV to 4 MeV for gamma photon. The third part of this work includes the dosimetry of Auger electron emitters both at cellular and sub-cellular levels, which has been determined by employing Geant4-DNA physics model – a track structure code. In order to account for non-uniform activity distribution due to the variation in the radio-pharmaceutical pharmacokinetics in both normal and cancerous tissues, voxel S-values have been estimated for 0.01, 0.1, 3 and 6 mm voxel sizes, considering cubical geometry of different tissue composition. For non-uniform dose profiles Dose point kernels have been estimated for 10 keV, 15 keV, 50 keV, 100 keV, 1 MeV and 4 MeV energies for mono-energetic electrons in water, lung, bone and air materials. The assumption of homogenous and uniform distribution of activity throughout the cell can lead to a large overestimation or underestimation of nuclear average dose rate. For nucleus uptake only, the dose rate to the nucleus will be under-estimated by ~90% when compare to the dose rate value for whole nucleus. On the other hand, the dose rate to the nucleus will be over-estimated by 27% and 12%, for radionuclide’s distributed within cytoplasm and cell surface respectively.