Romanian Biotechnological Letters Vol. 22, No. 4, 2017 Copyright © 2017 University of Bucharest Printed in Romania. All rights reserved ORIGINAL PAPER Romanian Biotechnological Letters, Vol. 22, No. 4, 2017 12802 Comparison of models for genetic analysis of traits from performance test of gilts using restricted maximum likelihood Received for publication, November 11, 2015 Accepted, April 22, 2016 DRAGOMIR LUKAČ 1 *, BRANISLAV MIŠČEVIĆ 2 , TIBOR KÖNYVES 2 , NIKOLA PUVAČA 3 , NATALIJA DŽINIĆ 4 , OLIVERA ĐURAGIĆ 5 1 University of Novi Sad, Faculty of Agriculture, Trg Dositeja Obradovića 8, 21000 Novi Sad, Serbia 2 Faculty of Biofarming, „John Naisbitt“ University Belgrade, 24300 BačkaTopola, Marsala Tita 39. Serbia 3 Patent Co., 24211 Mišićevo, Vlade Ćetkovića 1A, Serbia 4 University of Novi Sad, Faculty of Technology, Bulevar cara Lazara 1, 21000 Novi Sad, Serbia 5 University of Novi Sad, Institute of Food Technology, 21000 Novi Sad, Bulevar cara Lazara 1, Serbia *Address correspondence to: University of Novi Sad, Faculty of Agriculture, Trg Dositeja Obradovića 8, 21000 Novi Sad, Serbia. Tel.: +381 643526184; Email: dragomirlukac@gmail.com Abstract The objectives of this research was to compare estimates of variance components using different animal models in order to determine whether simpler models produce estimates similar to those produced by more complex alternatives, and to determine the most suitable mixed model for estimating genetic parameters and genetic trends for traits in performance test of gilts using REML. Four mixed model was constructed: M1 (Random effect of Animal and fixed effects of farm, year, season and breed), M1 (M1 + regression influence of body weight at the end of performance test), M3 (M1 + regression influence of duration of performance test), and M4 (M2 + regression influence of body weight at the end of performance test). In examined mixed models, information criteria was lowest in model M4, which would be the most adequate model, while model M1 had largest information criteria, which suggested that this model was not suitable model for evaluation of genetic parameters. Strong negative genetic and phenotypic correlation is showed between MP and BFT1 and between MP and BFT2, while between AG and BFT1 and BFT2 established positive and negative genetic and phenotypic correlation. Between BFT1 and BFT2 was determined strong positive genetics and phenotypic correlation. Heritability traits in four models were at medium to high degree of heritability. The resulting genetic trends were different between models, with relatively high coefficients of determination (R2). Keywords:REML, genetics analysis, models, performance test of gilts 1. Introduction Genetic improvement in swine production is based on the selection of the best and genetically superior animals in the population which takes in account economically valuable traits. This criteria will be used in further reproduction of parents in order for them to be the next progeny generation with purpose of achieving the maximum genetic progress with application of basic principles of quantitative genetics. Without a quality breeding stock, it is not possible to perform a good remount on the farm and to produce sows with high quality with the aim to follow the increased production demands in this branch of livestock production. For this reason, special attention have to be paid to selection and breeding gilts on