In a prospective investigation, 99 very preterm infants (gestational age (GA) 24 32 weeks, birthweight 560-2,255 g) were studied during the first 4 weeks of life. The infants were divided into two groups: infants born extremely early (GA <28 weeks, n = 20) and infants of GA 28 - 32 weeks; the groups were then subdivided into critically ill or not. Diagnostic blood sampling and blood transfusion events were recorded. In total, 1905 blood samples (5,253 analysis) were performed, corresponding to 0.7 samples (1.9 analysis) per day per infant. The highest frequencies were found during the first week, in infants with extremely low GA and in critically ill infants. The mean blood loss and transfusion volume values were 13.6 ml/kg and 6.3 ml/kg, respectively. In total, 19 infants (19%) received 34 transfusions corresponding to 0.3 transfusions per infant. Thirteen out of 20 infants of extremely low GA received 28 blood transfusions, corresponding to 27.0 ml/kg of blood on average during the study period. Four developed late anaemia; thus, in total, 14 (70%) of the infants born extremely early received 35 transfusions during the first 3 months of life, corresponding to a total mean of 34.8 ml/kg. For the extremely preterm infants a significant correlation between sampled and transfused blood volume was found (mean 37.1 and 33.3 ml/kg, respectively, r = + 0.71, p = 0.0003). The most frequently requested analyses were glucose, sodium and potassium. Few blood gas analyses were requested (1.9/ infant). No blood losses attributable to excessive generous sampling were detected. The results show an acceptable low frequency of sampling and transfusion events for infants of GA 28-32 weeks. The study emphasizes the necessity of thorough reflection and monitoring of blood losses when ordering blood sampling in extremely preterm, critically ill infants.