A randomized, double-blind, controlled trial was conducted to determine whether vancomycin added to parenteral alimentation solution given via a central venous catheter would decrease the incidence of catheter-related coagulase-negative staphylococcal sepsis. Seventy infants with a central venous catheter (CVC) in place were randomly selected to receive total parenteral nutrition--either the standard solution or a solution containing 25 micrograms of vancomycin per milliliter. Catheter-related sepsis was defined as the isolation of the same bacterial species from specimens of both peripheral and CVC blood with the concentration of bacteria at least tenfold greater in the specimen obtained from the CVC. Specimens from the CVCs were cultured on removal of the catheters to determine colonization. The colonization of catheters by coagulase-negative staphylococci was reduced from 40% to 22% (p = 0.03) in the vancomycin group; catheter-related sepsis was reduced from 15% to no cases (p = 0.004). Fewer infants required CVC reinsertion in the vancomycin-treated group (p = 0.02), who also regained birth weight earlier (13.4 vs 17.1 days (p = 0.014)). Adverse effects of vancomycin infusion were not observed. We conclude that vancomycin added to the solution used for total parenteral nutrition effectively reduces catheter-related sepsis in the neonatal intensive care unit and offers other potential benefits such as the need for fewer catheters and earlier weight gain. However, we do not recommend widespread implementation of this technique until there are data regarding the emergence of vancomycin-resistant organisms.