Many studies suggest that breast feeding confers developmental and intellectual advantages on children. In a recent study, however, no association was found between breast feeding and intelligence in adult life after adjustment for other variables, and the use of pacifier in infancy was the most important predictor of intelligence. We analysed the associations between breast-feeding duration, pacifier use and suspected developmental delay at 12 months of age in a birth cohort in Pelotas, southern Brazil. All 5304 hospital births occurring during 1993 were studied and a sample was followed up at 1, 3, 6 and 12 months of age. Breast-feeding practices and use of pacifiers were assessed at each visit, as well as suspected developmental delay, measured by the Denver II test. The prevalence of developmental delay was analysed, through logistic regression, according to breast-feeding status and pacifier use, accounting for the possible confounding effect of other variables. The prevalence of suspected developmental delay at 12 months was 34.1%, being slightly higher among children who used pacifiers at 6 months than among non-users (35.3% and 28.7% respectively). There was a marked negative association between breast-feeding duration and developmental delay, with children breast fed for 9 months or more presenting significantly less suspected developmental delay (25.5%) than those breast fed for less than 1 month (42.4%). The effects of multiple variables were tested, and only high parity, smoking during pregnancy, birthweight, gestational age, pacifier use and breast-feeding duration remained significantly associated with suspected developmental delay. The effect of pacifier use, however, disappeared after adjusting for breast-feeding duration, suggesting that breast feeding, and not pacifiers, affects child development.