Distributed generation (DG) sources are becoming more popular nowadays because of their technoeconomic and environmental benefits. These benefits are maximized only when the DG sources are properly planned and installed in the distribution networks. Several studies have been carried out in this field considering various objectives and constraints, for single as well as multiple DG placement and sizing. In this paper, an optimization methodology based on particle swarm optimization is used for the optimal planning of multiple DG sources in a meshed distribution network. The solution consists of the possible DG locations, DG capacities, and its operating power factor. The objectives considered are the improvement of reliability indices namely the system average interruption frequency index and system average interruption duration index, reduction of real power loss, and voltage profile improvement. Unlike other works, reliability evaluation is done considering many realistic constraints and it is used as one of the criteria for DG planning. Reliability indices are evaluated using the encoded Markov cut set algorithm. The real power loss and the voltage index are obtained by means of a simple load flow technique. The stochastic nature of the renewable DG output and the load are taken into account during the reliability evaluation along with islanding probability. The optimal planning of DG sources is done for bus 4 of the Roy billinton test system consisting of seven feeders, 102 nodes, and 106 components. Optimal planning for solar based DG sources is done considering three different load models namely residential, commercial, and mixed load. © 2019 Author(s).