Objectives To formulate a multilevel model for the longitudinal study of global weight gain rate in the Kangaroo-Mother Care Program. Methodology A multilevel model is used, taking time and some infant and mother variables in the first level. Measures associated with the global weight gain rate are considered for the second level. Results At 5 % significance, the intercept and the parameter associated to the time (days after birth) squared variable have a significant effect on the global rate of weight gain. The slope for the time variable is significant at the 6 % level. Also, under this model, we find that the gender delay in intrauterine growth of the infant, height and perimeter do not have a significant effect on the infant's global rate of weight gain. The weight in the various controls performed is significant at the 5 % level. There are differences between the two infant's gestational classifications (1: 26 to 40 weeks; 2: greater than 40 weeks), and between the infant's weight classifications of the categories two with one, and three with one (1: 2500 gm or more; 2: between 1500 and 2499 gm; 3: less than 1500 gm). Conclusion For each additional appointment, the average weight gain rate of the infant decreases. However, the infant keeps gaining weight, but not at the same rate as in the initial appointments.