This paper first presents a brief review of existing computation techniques and challenges in distribution system state estimation. Then, the impacts of measurement points, network configuration, load model, distributed generation, real-time measurement data, and computational parameters on the state estimation of distribution grids are investigated. Simulations are carried out on an example utility distribution network which is fed by a large high voltage transmission network at multiple points. The results show that greater observability of downstream distribution network enhances accurate estimation of network states. Similarly, load models and distributed generations have significant impact on state estimation. The results also show that the accuracy of state estimation depends on the choice of computational parameters.