The article examines the use of a metaheuristic method – genetic algorithms – for evaluating a model of routes for raw milk collection. A model was implemented based on real data, collected through fieldwork, following the method of the «traveling agent’s problem», using the toolbox of Matlab® genetic algorithm. The results show that the routes obtained with the implementation of the genetic algorithm are feasible in terms of time and visited nodes, demonstrating the potential of this tool. The costs obtained by this method differ from the current methods by about 3%, which is within the range reported in the literature.