This paper presents a methodology for optimal multistage planning of Distributed Generation to improve the performance of the distribution system. The methodology considers the optimal placement and sizing of the distributed generation to minimize network losses using Particle Swarm Optimization algorithm for different scenarios and stages. If is necessary, in each stage an optimal placement and sizing were determinate, the solution proposed must not violate the limits of loading and voltage in the distribution system. The proposed methodology was implemented in a collaborative work between Python® and DIgSILENT PowerFactory®. The methodology was validated on a modified IEEE 13 test feeder in four scenarios, consider regional demand growth in the Colombian Caribbean Region. The results showed the effectiveness of the methodology with losses reduction approximately of 84.65% in scenario 2 and 29.78% in scenario 4. Moreover, the methodology improved the voltage profile with demand growth of 3.68% per stage.