Abstract: The objective of this paper is to provide an introduction to generalized linear mixed models(GLMM), and to compare the efficiency of heritability estimate for discrete traits using three methods (LM, TM, GLMM). Binary response traits are simulated in this study, and the design is mixed family of full-half sib. The results showed that generalized linear mixed model have a great advantage in heritability estimation, and have nice accuracy for discrete traits. The results of estimation heritability using GLMM method are higher than those using LM method in all different parameter case. In addition, the accuracy of estimation depended on the true heritability and incidence of the trait directly.