The integration of distribution generation (DG) in distribution networks with improper planning adversely influences the quality of the electrical networks. Conventionally, the outputs from the intermittent DGs, such as solar photovoltaic (PV) plants, are assumed dispatchable. The intermittency of solar irradiance on the outputs of the PV modules has been ignored in most studies on the sizing and placement of DGs. By looking at this problem, this paper presents the sizing and placement of a distributed solar photovoltaic plant (DSPP) by using time series historical weather data. To predict the output from the PV modules, 15 years of solar data were modeled with the aid of a beta probability density function. The total energy loss index was formulated as the main objective function, and the optimization problem was solved by mixed integer optimization by using genetic algorithm. By adopting a time-varying commercial load, the proposed algorithm was applied on IEEE 33 bus and IEEE 69 bus distribution networks. The numerical studies on the two distribution networks show the advantages of the proposed approach for minimizing the total energy losses and improving the bus voltage profiles. It was revealed that up to 38 of the total energy losses in distribution networks could be reduced at sites with solar insolation of 5.65 peaks sun hours. In contrast to existing methods, planning for DGs by using weather data provided more realistic results for DSPP in distribution networks. © 2018 Author(s).