A NUMBER OF questions remained unanswered by the empirical success of antimicrobial prophylaxis for neurosurgical patients at The Mount Sinai Hospital during a 15-year period. Vancomycin (1 g intravenously) and tobramycin (80 mg intramuscularly) were administered in the induction room. Streptomycin (50 mg) was mixed into each liter of saline used to irrigate the surgical incision. A series of 45 consecutive clean neurosurgical operations were investigated. The potential sources of random contamination of the surgical wound that were studied included the following: 1) the patient's skin; 2) the flora of the skin and nares of the operating team; 3) the surgical apparel; 4) the surgeons' gloves; and 5) the airborne organisms in the operating theater. No wound infections were documented during a 4-month period between June and September of 1991. A remarkable 98% of the intraoperative cultures of the surgical wounds were free of bacteria. Positive cultures of glove imprints were found in 29% of the operations, and the bacterial source was traced to four different surgeons in four operations (9%). The surgeons' gloves were also a source of potential pathogens (Staphylococcus aureus) in two instances, but the bacterial species were also recovered from cultures of the environment. Based on individual biotyping of bacteria and antibiotic susceptibility testing, no consistent source or pattern could be uncovered for the bacteria in the surgical wound or the operating room air. The regimen of parenteral vancomycin and tobramycin and topical streptomycin was justified by the data that 36% of the operations were at potential risk for methicillin-resistant staphylococci or Gram negative bacterial species cultured from the patient's skin (5 cases), contaminated surgeons' gloves (1 case), the operating room environment (8 cases), or both the contaminated surgeons' gloves and the operating room environment (2 cases).