objective: to assess the incidence of infantile colic and its determinants. methotds: between may and july 1999, the research team performed daily visits to the three main maternities of pelotas. all women were interviewed after the delivery and their infants were targeted for follow-up at three months. colic was defined according to wessel. the following risk factors were evaluated: social class, mother's educational level, parents' age, parents' divorce, changes in the parents' relationship during pregnancy, type of delivery, reproductive history, quality of prenatal care, previous abortion, previous stillbirths, previous newborn with health problems, gender, and infant's feeding pattern. chi-square test was used to compare proportions and conditional logistic regression was applied in the multivariate model. results: we managed to follow 1,086 of the 1,195 infants. at the three months visit, 80.1% of the mothers reported that their infants had had colic. on the other hand, only 16.3% was considered as having colic according to wessel criteria. mother's educational level, father's age, type of delivery and breast-feeding duration remained in the multivariate model. even after controlling for possible-confounding factors, the chances of having colic was 1.86 times higher among non-breastfed infants than among breastfed infants. conclusion: most mothers misclassify the occurrence of colic. breastfeeding is the main protective factor.