BACKGROUND Imipenem-cilastatin (referred to hereafter as imipenem) is administered at different doses as monotherapy or with other agents. Limited comparisons of the alternatives exist. The authors compared the efficacy and safety of several imipenem-containing regimens (ICRs) to determine the appropriate dose and the need for combination therapy. METHODS Between 1985 and 1994, febrile neutropenic patients were given ICRs according to the same methodology on prospective trials at a referral cancer center. The ICRs were high dose imipenem (HIP), high dose imipenem and amikacin (HIPA), high dose imipenem and vancomycin (HIPV), and low dose imipenem and vancomycin (LIPV). RESULTS The overall response rates were comparable (70-77%). There was a univariate trend toward better response among patients with pneumonia and documented infections with unidentified organisms who received HIPV versus LIPV (P=0.06), as well as a significantly better response among patients with gram positive infections who received HIPV versus HIP (P=0.02) and HIPA (P=0.002). HIPV was a more effective treatment for documented infections with identified organisms (P=0.05) and bloodstream infections (P=0.04) than HIP; there was a univariate trend toward better response among patients infected with gram negative organisms who received HIPA versus HIP (P=0.12). Multivariate adjustment for baseline and prognostic factors did not reveal a relative advantage for any regimen. No differences in overall toxicities were observed between HIPV and LIPV. CONCLUSIONS Imipenem monotherapy is adequate treatment for most febrile neutropenic cancer patients. Low dose imipenem could be effective and safe in uncomplicated cases without pneumonia. Further studies are needed to establish the usefulness of low dose imipenem in this context.