We developed a decision rule to define a group of newborns eligible for discontinuation of antibiotic treatment 24 hours after initiation. The decision rule is based on two clinical assessments, three demographic variables, four maternal risk factors and six laboratory studies. The rule was created by retrospective analysis of all admissions to rule out sepsis at 10 Kaiser Permanente Northern California Region nurseries during a 2-month period in 1990. Of the 214 study subjects, 11 (5%) had a positive bacterial or viral culture result. No deaths occurred, but three babies who had negative culture results experienced late deterioration (after 24 hours in the special care nursery). After 24 hours of antibiotic treatment, 133 (67%) of the 199 babies who had negative culture results were asymptomatic, 36 (18%) had clear signs of illness and 30 (15%) had questionable signs of illness. The rule correctly identified all babies with positive culture results as well as all persistently symptomatic babies with negative culture results. The rule also identified 57 low-risk babies eligible for only 24 hours of antibiotic treatment. We tested the rule on a second data set. The rule identified 19 of 20 newborns with positive culture results in this independent group. The 20th baby was not identified because a required test (complete blood count at 12 to 24 hours of age) was not done. Under controlled clinical conditions, this baby could have returned to the nursery to complete antibiotic treatment. We concluded that early discontinuation of antibiotic treatment in selected newborns is safe and could lead to a 1 to 2% decrease in special care nursery days in our 10 hospitals.