A prospective, randomized study to evaluate the effectiveness of a continuous low-dose vancomycin infusion to prevent nosocomial gram-positive bacteremia was initiated within the first 2 weeks of life in neonates weighing < 1500 gm. Seventy-one infants received constant infusion of vancomycin (25 micrograms/ml) mixed with their total parenteral nutrition solution; 70 infants served as control subjects. The groups were clinically similar in birth weight, estimated gestational age, and severity of illness. Administration of vancomycin was begun at a mean age of 5.4 +/- 2.9 days. Infants had mean serum vancomycin concentrations of 2.4 micrograms/ml, and received vancomycin for a mean of 11 +/- 7 days. No vancomycin-resistant organisms were detected in surveillance cultures during the 2-year study period. Twenty-four of seventy control infants, in comparison with 1 of 71 infants receiving vancomycin, had gram-positive bacteremia (p < 0.001). The addition of a low dose of vancomycin to alimentation fluids virtually eliminated the incidence of gram-positive bacteremia in an at-risk population of very low birth weight infants. However, the widespread use of vancomycin in total parenteral nutrition solution is not recommended until better data on the emergence of vancomycin-resistant organisms are available.