Distributed generation (DG) connected to distribution networks affects the currents or power flows in the networks; thus, node voltages that are strongly related to power flows also are changed. As the voltage must be within permitted limits to comply with utility and customer requirements, this effect should be assessed prior to DG connection. At the distribution level there are a number of attributes that can influence the voltage profiles; therefore, this assessment requires detailed distribution operation models. To provide reliable voltage evaluation results, a new probabilistic methodology is proposed in this paper. The distribution system operation uncertainties including daily time varying load, stochastic DG power production, network configuration, and voltage control devices operation are all taken into account. A probabilistic load flow method based on an efficient algorithm is used to handle uncertainties in loads and DG output power and voltage control devices operation. Accordingly, a probabilistic network reconfiguration model is employed to take stochastic network configurations for service restoration or load balancing into consideration. The proposed probabilistic method provides a better knowledge of the voltage effects and can be used for evaluating the level of DG that might be accepted on a distribution network. The models developed can also be used for assessing impact of voltage mitigation equipment on distribution operating performance. Test results of a modified IEEE feeder test system demonstrate the performance of the proposed method.