The aim of this study was to evaluate the use of different mathematical models to describe the growth of lambs during the pre-weaning phase and the effect of sex on their growth curve. Data from 51 lambs (26 males and 25 females) with a genetic constitution of at least 50% Santa Inês breed were used. Lamb weight was monitored during the entire pre-weaning phase (91 days) by weighing the animals weekly on a scale suitable for sheep, to draw the body-weight growth curve. The adequacy of the models was evaluated by the coefficient of determination, simultaneous F-test for parameter identity, concordance correlation coefficient, root mean square error of prediction and decomposition of the mean square error of prediction. Pairwise mean square error of prediction analysis and delta Akaike information criterion were used to compare the models in terms of accuracy and precision. The evaluation of the adequacy of the tested models showed that the data predicted by the linear, quadratic and exponential models were similar (P > 0.05) to the observed data. In contrast, the monomolecular model generated predictions that could not be accepted as true. The linear and quadratic models showed better accuracy and precision. The first-degree linear model should be preferred over the other studied models to describe growth of sheep in the preweaning phase. Males and females exhibited distinct growth patterns.