A review of the duration of antibiotic courses in our Neonatal Intensive Care Unit (N.I.C.U.) has shown that in a significant number of cases there was non compliance with the antibiotic policy. An audit of neonatal sepsis and antibiotic usage over a six month period was performed in the N.I.C.U. at the Rotunda Hospital. Three hundred and forty-nine of the 3,163 infants born during this time were admitted to the N.I.C.U. One hundred and fifty-two infants had one or more episodes of suspected sepsis. In 168 instances this preceded a course of antibiotic therapy lasting longer than 48 hours. In 56 (33%) the infant was clinically septic and in 16 (9.5%) of those there was bacteriological proven sepsis. There were 112 episodes (60%) where there was no strong clinical evidence of sepsis and a negative investigation for sepsis where antibiotics were continued for more than 48 hours. The possible reasons why antibiotics were not stopped include the non specific nature of signs of sepsis in neonates, physicians' reluctance to discontinue antibiotics once started, and the logistic difficulty of obtaining routine culture results at weekends. The study emphasizes the need for regular surveillance of antibiotic usage in a N.I.C.U.