Data comprising68,523 test day milk yield of 8,536 cows of the Holstein breed, daughters of537 sires, distributed in 266 herds, calving from 1996 to 2001, were used tocompare random regression models, for estimating variance. Test day records(TD) were analyzed by different random regression models regarding the functionused to describe the trajectory of the lactation curve of the animals. Legendreorthogonal polynomials function of second, third and fourth order were used.The random regression models included the effects of herd-month-year of thecontrol, genetic group of the animals; the frequency of the daily milk; regressioncoefficients for each class of age-season (in order to describe the fixed partof the lactation curve) and random regression coefficients related to the directgenetic and the permanent environmental effects. The heritability estimatesobtained using the random regression models ranged from 0.122 to 0.291. Therandom regression model which used the fourth order Legendre polynomials wasthe model which better described the genetic variation of the milk yield, accordingto AIC test.
