Objective: To compare a short course of antibiotics (48 to 96 hours) and a standard course of antibiotics (7 days) for probable neonatal sepsis. Design: Randomized, controlled, open-labeled trial with blocking and stratification according to birth weight. Setting: Tertiary care, referral, teaching hospital in Northern India. Participants: Neonates >30 wks gestation and >1000 g at birth, with probable sepsis (clinical signs of sepsis, raised C-reactive protein) were enrolled. Babies with major malformations, severe birth asphyxia, meningitis, bone or joint or deep-seated infection, those who were already on antibiotics, and those undergoing surgery were excluded. Neonates, who had clinically remitted on antibiotic therapy - by the time a sterile blood culture report was received - were randomized. Intervention: In the intervention arm, antibiotics were stopped after the 48-hour culture was reported sterile. In the control arm, antibiotics were continued to a total of 7 days. Main outcome measure: "Treatment failure" defined as reappearance of signs suggestive of sepsis within 15 days of stopping antibiotics, supported by laboratory evidence and adjudicated by a blinded expert committee. Results: 52 neonates were randomized to receive a short course or 7-day course (n=26 each). Baseline variables were balanced in the 2 groups. There was no significant difference in the treatment failures between the 2 groups (3 babies in the 7-day group vs none in short course group, P=0.23). Conclusion: No difference in the treatment failure rates could be identified between short course and 7-day groups among neonates >30 weeks and >1000 grams with probable sepsis.