This work presents the different diversity measures found in the literature for determining whether a set of classifiers is sufficiently diverse, an aspect of great importance in the design of multi-classifier systems. It also introduces a model for building multi-classifier systems using the Genetic Algorithms metaheuristic in order to achieve the highest possible accuracy while maximizing diversity among classifiers. In addition, several approaches for combining diversity measures are described. Finally, two experiments are discussed in which both the individual behavior of the diversity measures and the results obtained from their combinations are analyzed.