The Medium Term Development Frame Work (MTDF) 2005-10 by Planning Commission Government of Pakistan states the policy for power sector in which it puts a greater emphasis on nuclear power resources by increasing its share from currently 425 MW to 8800MW by 2030. With the increase of nuclear share in the overall national energy mix, a corresponding environmental impact and nuclear safety analysis are equally important. These are usually taken care of by Primary Safety Analysis Report (PSAR) of a proposed nuclear power plant. The PSAR of any proposed nuclear power plant involves the assessment of a hypothetical accidental release of radionuclides in the atmosphere as set forth by US-NRC and PNRA such as those given as criteria for preparation and evaluation of radiological emergency plans and preparedness (10CFR100, PAK/910). Modeling atmospheric dispersion (both transport and diffusion) is the first step of such assessments. The objective of this work is to determine a more precise modeling methodology that can better predict the radiological consequences in terms of radionuclide concentration and doses compared to Gaussian dispersion approach that is based on assumptions such as uniform turbulence, flat topography and non-variant wind speed with time and space. The research goal was achieved by developing two broad strategies on the basis of Lagrangian approach. The first strategy is an effort to provide a simple answer to the complex problem. This methodology makes use of empirical parameterization of meteorology which serves as input for dispersion calculations by Lagrangian Stochastic Particle Model (LSPM). But the beauty of approach is its capability to capture complex atmospheric phenomenon like wind directional shear. This approach was used to study hypothetical accidental release of radionuclides in nocturnal atmosphere which generates maximum wind directional shear. The results of dispersion in terms of dispersion coefficients were in good comparison with that of experimental findings in the available literature. The resulting ground level concentrations of radio-nuclides and radiological dose contours were also compared with those based on approach analogous to Gaussian Plume Model (GPM). The exercise proved that how misleading results would be if we ignore wind directional shear in stable atmosphere. The second approach is based on a state of the art solution. It involves the coupling of an Eulerian meteorological model (RAMS) with LSPM. The meteorological model is responsible to provide meteorological input to LSPM at each grid point and at each time step. This computational technique was used to simulate a hypothetical accident at a proposed site for Nuclear Power Plant. The meteorological output of the modeling system was compared with observed values. The comparison proved the efficacy and reliance of the approach. This computationally intensive but effective strategy is quite capable of supporting a real time decision making system for tackling nuclear emergency.