BACKGROUND AND OBJECTIVES: Benefits of adopting restrictive guidelines for erythrocyte transfusions are still controversial. The objective of this study was to verify if a very strict guideline could reduce erythrocyte transfusions in preterm infants without adverse outcomes. MATERIALS AND METHODS: Two prospective cohorts of neonates with gestational age < 37 weeks and birth weight < 1500 g were studied. Neonates born in Period 1 were submitted to a strict guideline for erythrocyte transfusions. In Period 2, a new stricter protocol was introduced. Infants of both periods were compared regarding number of transfusions and clinical outcome. RESULTS: The median number of transfusions decreased from 2 (1 to 14) in Period 1 to 1 (1-9), P = 0.001, in Period 2. The linear regression multivariate analysis showed that the implementation of the stricter guideline was associated with a reduction in the number of transfusions received by patients by 0.55 (95% confidence interval: -0.08; -1.02) units/patients. Number of apnea episodes, weight at 28 days of life and days of hospital stay were similar in both periods. Intra-hospital death was lower in Period 2. CONCLUSION: A very strict guideline reduced the number of erythrocyte transfusions in preterm infants, without threatening their clinical course.