Mathematical models to predict the dry matter intake (DMI) of feedlot Santa Ines rams were developed and evaluated in this study. The available database had 100 experimental units from 13 studies. Study effect was integrated and random effects of their interactions as components of a hybrid model. The independent variables were initially adjusted to a model which included fixed effects for y-intercept and slope and random effects in y-intercept and slope study, using unstructured covariance model (e.g.: UN-unstructured). Study effect on database was verified, and then a meta-analysis procedure to develop DMI prediction equations was performed. For validation and comparisons between existing prediction equations in the national and international literature, independent data from one survey with 21 animals were used. Validation methods of the observed and predicted DMI were based on linear regression model adjustment of the observed values over predicted values. The following variables: average live weight (ALW), metabolic live weight (MLW0.75), average daily gain (ADG) and average daily gain2 (ADGsq) presented positive correlation with DMI. In contrast diet concentrate level showed a negative correlation. Among eight models examined, the following resulting equation [DMI (g/day) = 238.74 ± 114.56 (0.0398) + 31.3574 ± 4.2737 (<0.0001) × MLW + 1.2623 ±0.2128 (<0.0001) × ADG − 5.1837 ± 0.7448 (<0.0001) × CON] has been found as the best fit model to predict DMI in feedlot Santa Ines rams.
