Two hundred twenty neonates with suspected early onset sepsis were prospectively studied to evaluate the ability of a sepsis screen to discriminate infected from noninfected newborn infants. A positive sepsis screen consisted of positive findings in two or more of the following tests: total white blood cell count; immature/total neutrophil ratio; C-reactive protein; micro-erythrocyte sedimentation rate; or plasma fibronectin. For proved sepsis a four-part screen excluding fibronectin yielded a sensitivity of 100%, specificity of 83%, positive predictive value of 27% and negative predictive value of 100%. In contrast the sensitivity of white blood cell count and immature/total neutrophil ratio was only 46%. Adding fibronectin to the four-part screen provided equal sensitivity and negative predictive value but decreased specificity and positive predictive value. While plasma fibronectin may play an important role in the pathogenesis of neonatal sepsis, it is not useful as a marker for infection. The screens did not identify preterm infants with late onset nosocomial sepsis. Although clinical judgment should be the primary factor in the decision to institute antibiotic therapy, a simple four-part sepsis screen provides valuable presumptive information for excluding the diagnosis of early onset neonatal sepsis.