Aim: To study whether a simple targeted intervention could reduce unwarranted antibiotic treatment in near-term and term neonates with suspected, but not confirmed early-onset sepsis. Methods: A quality improvement initiative in three Norwegian neonatal intensive care units. The intervention included an inter-hospital clinical practice guideline for discontinuing antibiotics after 36-48 hours if sepsis was no longer suspected and blood cultures were negative in neonates ≥ 34+0 weeks of gestation. Two units used procalcitonin in decision-making. We compared data 12-14 months before and after guideline implementation. The results are presented as median with interquartile ranges. Results: A total of 284 infants (2.5% of all births ≥ 34+0 weeks of gestation) received antibiotics before and 195 (1.8%) after guideline implementation (P = .0018). The two units that used procalcitonin discontinued antibiotics earlier after guideline implementation than the unit without procalcitonin. Neonates not diagnosed with sepsis were treated 49 (31-84) hours before and 48 (36-72) hours after guideline implementation (P = .68). In all infants, including those diagnosed with sepsis, antibiotic treatment duration was reduced from 108 (60-144) to 96 (48-120) hours (P = .013). Conclusion: Antibiotic treatment duration for suspected, but not confirmed earlyonset sepsis did not change. However, treatment duration for all infants and the proportion of infants commenced on antibiotics were reduced.