The optimization of economic load dispatch (ELD) is one of the oldest and most important tasks in power plant management. The objective of this paper was to analyze a new solution of the old problem of the ELD optimization by differential evolution (DE) including turning off the most inefficient generators. The criteria of incremental fuel costs and losses are used to determine the best parameters of active power of each i th generator unit, ensuring that the demand and total losses are equal to the total generated power but minimizing the total cost of fuel. Materials and methods have been developed to solve the ELD, including lambda iteration method, gradient method, Newton’s method and so on. The results found for this case study, with the application of DE, were outstanding having a reduction of 19.88 % in the total fuel cost, comparing to classical methods that distribute the generation of power among all generators, including the least efficient ones. This method improves not only the efficiency of generation but also of the power plant generation planning.