Antimicrobial prophylaxis against gram-positive bacteremia (GPB) following BMT may prevent infections but promote antimicrobial resistance. In a sequential cohort study involving 289 consecutive BMT recipients we compared three protocols for prevention of GPB (vancomycin prophylaxis, penicillin/cefazolin prophylaxis, and no specific GPB prophylaxis) with respect to incidence of GPB, mortality, and vancomycin use. GPB was associated with increased mortality (27% vs 15%; P = 0.02), but contributed to only five of 52 deaths in the study population, and only one of 15 subjects with viridans streptococcal bacteremia developed fatal septic shock. Vancomycin prophylaxis reduced the incidence of GPB (11%) compared to penicillin/cefazolin (27%) or no prophylaxis (40%) (all P < 0.03), but did not significantly reduce mortality. The incidence of fungemia, gram-negative bacteremia, and infection-associated mortality was unaffected by GPB prophylaxis. Vancomycin use was substantially greater in the vancomycin prophylaxis group. We conclude that in comparison with vancomycin prophylaxis, BMT support regimens that do not include vancomycin prophylaxis allow reduced overall vancomycin use without an apparent increase in early post-BMT mortality, despite the greater associated frequency of GPB.