Composite bovine data was analyzed with the objective of evaluating the effect of the epistasis parameter in the models of genetic evaluation. The analyzed characteristics were weight at 205 (W205) and 390 days (P390), and scrotal circumference at 390 days (SC390). The analysis were done by the maximum likelihood method, considering two models: model 1, which included as covariates the direct and maternal additive effects, and non-additive of the heterozygosis for the direct and total maternal, and model 2, which also considered the direct epistasis direct. The Akaike Information Criteria (AIC) and the Bayesiano of Schwartz Information Criteria (BIC) were used for the comparison of the models and the test of ratio of likelihood. The inclusion of the epistasis effects on the model of the genetic evaluation did not alter much the estimation of the genetic additive (co)variances components and, consequently the heritability. However, it was significantly superior by the likelihood ratio test for the studied characteristics. Through the BIC, model 2 was more adequate only for W205. For the genetic analysis of that population the model that considers the epitasis is the more adequate.
crossbreed; non-additive genetic effects; scrotal circumference; weights
