A prospective randomized study was conducted to determine the efficacy of imipenem-cilastatin (hereafter referred to as imipenem) (500 mg four times daily) versus combination therapy for febrile neutropenic patients receiving either no prophylaxis or ciprofloxacin for prevention of infections. Combination therapy consisted of gentamicin (80 mg every 8 h) plus either cefuroxime (1,500 mg every 8 h) or cephalothin (1,000 mg every 4 h) for suspected catheter-related infections. Ninety-four neutropenic fever episodes in 87 patients were evaluable for efficacy. The overall clinical rate of response to imipenem was significantly higher than that to combination therapy (91 versus 74%; P = 0.05). The difference in efficacy was most pronounced in patients with microbiologically documented infections (89 versus 53%; P = 0.025), which were predominantly caused by gram-positive bacteria. Differences in susceptibility may have caused the better rate of response to imipenem. Two of 29 gram-positive bacteria were imipenem resistant, whereas 10 were resistant to cephalothin and cefuroxime and 12 were resistant to gentamicin. No causative gram-negative bacterium and 24 gram-positive bacteria were isolated in 61 fever episodes with ciprofloxacin prophylaxis (oral). In contrast, nine causative gram-negative and five gram-positive bacteria were isolated in 33 episodes without prophylaxis. The difference in distribution proved to be statistically significant for gram-negative (P = 0.0001) as well as gram-positive (P = 0.025) bacteria, indicating that ciprofloxacin effectively prevented the occurrence of gram-negative bacteria and may have contributed to the relatively large number of gram-positive bacteria isolated. Empirical initial therapy with imipenem may be a valuable alternative to combination therapy for neutropenic fever episodes.