The estimation of the genetic merit for animals in a livestock population is carried out through genetic evaluations. Today, the method of choice for conducting genetic evaluations is the so-called single-step genomic BLUP (ssGBLUP), which combines phenotypes, pedigree, and genomic information in a single evaluation. Single-step GBLUP outputs the prediction of the genetic merit of the animals or estimated breeding values. The quality of these predictions is evaluated in terms of bias, inflation or dispersion, and accuracy. The objective of this dissertation was to perform cross-validation in terms of bias, dispersion, and accuracy on complex models and datasets using ssGBLUP.