To evaluate the interest of examination of the gastric aspirate (GA) as a contribution to decision making of initial antimicrobial therapy, all 3,989 neonates delivered in Orléans Maternity Hospital in 1990 have been studied. Microscopic examination: polymorphonuclear leukocytes (PMN) were, respectively, absent/present/abundant in 180/130/25 treated (T) and in 2,567/1,032/90 untreated (NT) newborns. PMN were demonstrated in 2/5 documented, 4/6 obvious, 2/2 suspected and 34/58 possible infections. Bacteria were, respectively, absent/present/abundant in 201/109/46 T newborns and in 2,722/877/56 NT newborns. Bacteria were demonstrated in 3/5 documented, 4/6 obvious, 2/2 suspected and 32/58 possible infections. Culture: the number of initially T and NT newborns was, respectively, for each bacterial species: Enterobacteriaceae 33/294, streptococci B (GBS), D and alpha-hemolytic 60/107, 12/110 and 18/70, Streptococcus pneumoniae 2/2, anaerobes 12/402, Listeria 3/0, Haemophilus species 4/4, Staphylococcus aureus 5/7, coagulase-negative staphylococci 25/1335, Lactobacillus 14/345, corynebacteria 10/196. In the 5 newborns with documented early onset meningitis and/or septicemia, 3 grew with GBS, Escherichia coli and Listeria. In the 4 newborns (2 meningitis and 2 urinary tract infections) with late-onset infection, no positive GA could be demonstrated. Although microscopic examination and cultures were statistically more frequently positive in T newborns, with variations dependent on the species of bacteria, more newborns without infection were colonized whatever the bacteria. Thus, bacteriological results of GA, if considered out of the clinical context, cannot be an argument for treatment.