The purpose of the study reported here was to investigate the impact of prophylaxis against gram-positive infections in patients undergoing high-dose chemotherapy and autologous bone marrow transplantation in a randomized trial. Forty-three patients undergoing high-dose chemotherapy with autologous bone marrow transplant were enrolled in a nonblinded randomized trial to receive or not to receive prophylaxis for gram-positive infections with 10(6) U of penicillin intravenously (i.v.) every 6 h (q6h) (if penicillin allergic, 750 mg of vancomycin i.v. q12h) in addition to standard antimicrobial prophylaxis with 400 mg of norfloxacin orally three times a day, 200 mg of fluconazole orally once a day, and 5 mg of acyclovir per kg of body weight i.v. q12h. The patients were being treated for germ cell cancer (n = 15), breast cancer (n = 16), Hodgkin's disease (n = 3), non-Hodgkin's lymphoma (n = 4), acute myeloid leukemia (n = 1), acute lymphoblastic leukemia (n = 1), and ovarian cancer (n = 3). The trial was stopped because of excess morbidity in the form of streptococcal septic shock in the group not receiving gram-positive prophylaxis. There were significantly fewer overall infections (10 versus 3; P = 0.016) and streptococcal infections (9 versus 1; P = 0.0078) in the group receiving gram-positive prophylaxis. There were no significant differences in the numbers of deaths, duration of broad-spectrum antibiotics, or incidence of neutropenic fever between the two groups. Prophylaxis for gram-positive infections with penicillin or vancomycin is effective in reducing the incidence of streptococcal infections in patients undergoing high-dose chemotherapy and autologous bone marrow transplant. However, this approach may carry a risk of fostering resistance among streptococci to penicillin or vancomycin.