We have studied all newborns admitted to our neonatal intensive care unit during 1993 and treated with intravenous antibiotics. Patient-files were examined for all available data at admission, focusing on factors predisposing for infection, symptoms, additional diagnoses, laboratory tests, bacteriology and antibacterial treatment. Antibiotics were given to 126 (28%) patients, of whom 90 were suspected of having an infection on admission. 57 of these were discharged with an infection-related diagnosis. 33 patients received prophylactic antibiotics, of whom three later developed infection. Retrospectively, 53 patients had proven or very probable infection. Fourteen patients tested blood culture positive. In our material the incidence of septicaemia was 0.45% of all newborn. Both the frequency of treatment and the incidence of septicaemia are consistent with the findings in earlier reports. We find that our material contains an unacceptably high frequency of false negative blood cultures. Recently published data show that the incidence of positive blood cultures is proportional to the amount of blood extracted.