Preterm neonates represent a category of patients with high transfusion needs. Ideally, red blood cells (RBC) transfusion should be tailored to the individual requirements of the single infant. However, despite the progress in neonatal transfusion medicine, many controversies still remain, and the decision on whether to transfuse or not is often made on empirical basis, with large variation in transfusion practices among neonatologists. Recently, a few clinical trials have been performed with the aim to compare the risk/benefit ratio of restrictive versus liberal transfusion criteria. Most of the studies failed to demonstrate significant differences in short-term outcomes, suggesting that the restrictive criteria may reduce the need for transfusion and the related side effects. Neurodevelopmental long-term outcome seemed more favorable in the liberal group at a first evaluation, especially for boys, and significantly better in the restrictive group at a later clinical investigation. Magnetic resonance imaging scans, performed at an average age of 12 years, showed that intracranial volume was substantially smaller in the liberal group compared with controls. When sex effects were evaluated, the girls in the liberal group had the most significant abnormalities. In conclusion, when preventive measures, as favoring cord clamping delay or cord milking, ensuring optimal nutrition, and minimizing phlebotomy losses, fail to avoid the need for transfusion, it is preferable to adopt restrictive criteria.