OBJECTIVE: To describe feeding practices and the risk factors for the mixed breastfeeding and early weaning in the neonatal period. METHODS: Cohort study, which we collected socioeconomic, demographic, health care and feeding data from 415 mother/child binomials born in four public maternity hospitals in Natal/Brazil. They were followed-up at 48 hours, 7 and 28 days after birth. The association was established using Pearson's Chi-square test and Poisson's regression, after adjusting it to other variables. RESULTS: The prevalence of mixed breastfeeding in the first 2 days was 47,2% and early weaning in 7 and 28 days was 8,4% and 16,2% in that order. The main reasons for mixed breastfeeding and early weaning were: colostrum deficiency (33.8%), difficulty in latching/sucking (23.5%) and "little milk" (70.0%). The use of formula/milk/porridge remained associated with maternal age ≤ 20 years (RR = 0.64; 95%CI: 0.47-0.86), age 20-29 years (RR = 0,70; 95%CI: 0,57-0,87), primiparity (RR = 1.37; 95%CI: 1.11-1.60) and cesarean delivery (RR = 1.20; 95%CI: 1.00-1.45) at 2 days; absence of paternal support (RR = 4.98; 95%CI: 2.54-9.79) and pacifier use (RR = 3.21; 95%CI: 1.63-6.32) at 7 days; and only pacifier use (RR = 2.48; 95%CI: 1.53-4.02) at 28 days. CONCLUSIONS: Early weaning was associated with maternal and health care factors, thus suggesting the need to readjust good practices and educational actions to achieve the exclusive offer to the maternal breast in the neonatal period.