Tools and algorithms are proposed that are useful for planning, designing, and operating a distribution network with a significant penetration of distributed generation (DG). In Task 1, a PV system simulation program is developed, which incorporates the most rigorous models for the calculation of insolation, module temperature, and DC and AC power output of a PV system. The effect of random inverter failures is incorporated in the model of a PV system, and a novel performance-derating coefficient is introduced. Furthermore, a novel inverter control algorithm is presented for systems with multiple inverters. The algorithm is designed to increase overall DC/AC conversion efficiency by selectively shutting down some of the inverters during periods of low insolation, thus forcing the remaining inverters to operate at higher efficiency. In Task 2, a procedure is developed to incorporate the uncertainties imposed by stochastic, renewable DG into the conventional tools for analysis of distribution systems. A clustering algorithm is proposed to reduce large input data sets that result from the interaction of stochastic processes that drive DG output with field measurements of feeder load profiles. In addition, a procedure is proposed to determine the boundary points of the original data set, which yield feeder extreme operating conditions. Finally, a Monte Carlo analysis using a reduced data set is presented, to determine the effects of deploying a large number of renewable DG systems on a distribution feeder. In Task 3, the reliability model of an asymmetric, three--phase, non-radial distribution feeder equipped with capacity-constrained DGs is developed and used to quantify the potential reliability improvements due to the intentional islanded operation of parts of the feeder. A procedure for finding optimal positions for DG and protection devices is presented using a custom-tailored adaptive genetic algorithm.