In our nursery, we identified neonates with positive urine latex particle agglutination (LPA) tests for group B streptococcus (GBS) antigen who did not have corroborating cultural evidence of infection. To investigate the mechanisms underlying these apparent false-positive reactions, we examined the urine LPA test in an unselected population of neonates suspected of sepsis. Urine specimens for LPA testing and culture and simultaneous perirectal cultures were obtained from 134 neonates with suspected sepsis. Six infants had positive blood cultures for GBS; four of the six were positive by LPA testing. An additional 20 infants had positive LPA tests but negative blood cultures; of these, 13 had mothers who received antibiotic treatment prior to delivery. Two infants with positive LPA results and negative blood cultures had bacteria isolated from urine cultures obtained in a nonsterile fashion (GBS, Escherichia coli). GBS was not isolated from perirectal swabs of infants with positive LPA tests and negative blood cultures. In conclusion, (i) a high proportion of neonates evaluated for sepsis gave positive LPA tests and negative blood cultures, (ii) local contamination of the perirectal skin or urinary tract with GBS was an unlikely source of false-positive LPA reactions, and (iii) maternal antibiotic pretreatment during labor may represent an important cause of apparent false-positive LPA reactions.