Prolonged antibiotic therapy is associated with antimicrobial resistance and increased mortality in preterm infants. We evaluated the impact of an automatic stop order (ASO) and C-reactive protein (CRP) on the duration of antibiotics and level of intervention in infants screened for early-onset sepsis who had negative cultures. We introduced an ASO for low-risk infants, then, consequently, for all infants treated for suspected sepsis. We subsequently introduced a single CRP measurement at 36 hours. Between 2011 and 2014, 4 time periods were studied, at baseline and after each intervention. The proportion of infants receiving ≤48 hours of antibiotics increased from 19% to 72.5% (P < .0001), whereas that of infants receiving avoidable doses (>48 hours and <5 days) fell from 50% to 0.8% (P < .0001). The use of an ASO decreased the proportion receiving avoidable doses from 26/92 (28.3%) to 9/293 (3.1%); P < .0001. There was a reduction in lumbar punctures performed, from 35% to 20%; P = .015.