To identify risk factors for discontinuing breastfeeding during an infant's first year of life. A cohort study recruited mothers in a hospital in São Leopoldo, Brazil, which mainly serves the low-income population. In order to obtain socioeconomic, environmental, and behavioral information, face-to-face interviews with mothers were conducted after birth, and when their infants were 6 and 12 months old. The duration of breastfeeding was investigated at 6 and 12 months, and recorded separately for each month. Depressive symptoms were assessed using the Beck Depression Inventory. The multivariate model for predicting the discontinuation of breastfeeding, adjusted Kaplan-Meier survival curves and Cox regression were used. Of the 360 participants, 201 (55.8%) discontinued breastfeeding within the first 12 months. A multivariate Cox regression model revealed that symptoms of maternal depression (low levels: RR = 1.59, 95% CI 1.02-2.47; moderate to severe: RR = 2.03, 95% CI 1.35-3.01), bottle feeding (RR = 2.07, 95% CI 1.31-3.28) and pacifier use in the first month of life (RR = 3.12, 95% CI 2.13-4.57) were independently associated with the outcomes after adjusting for confounders. Breastfeeding cessation rates were lower for children who did not use bottle feeding or a pacifier in the first month of life and for the children whose mothers presented with minimal depression. Early pacifier use and bottle feeding must be strongly discouraged to support long-term breastfeeding. In addition, screening maternal depression at a primary care service can be a step forward in promoting a longer duration of breastfeeding.