This open, comparative, randomized, multicentre equivalence study compared cefepime 2 g bd and imipenem-cilastatin 1 g tds (50 mg/kg/day) as empirical monotherapy for febrile episodes in a homogeneous cohort of cancer patients with short duration neutropenia following chemotherapy for solid tumour, lymphoma or myeloma. The study was conducted in 17 French anti-cancer centres in 1995 and 1996. Response to monotherapy was assessed 7 days after treatment and was based on resolution of fever and signs and symptoms, eradication of pathogens, absence of new infection, relapse, and death of infectious origin, without addition of other antibiotics. Patients were treated for a minimum of 4 days. Of the 400 episodes randomized, 344 (86%) were evaluable for efficacy. Patient characteristics were comparable between treatment groups. Success of monotherapy was observed in 79% of episodes with cefepime and 72% with imipenem-cilastatin (equivalence, P <0.0001). The response rate for microbiologically documented infections was 66% with cefepime and 61% with imipenem-cilastatin (bacteraemic episodes: 63% for cefepime; 44% for imipenem-cilastatin). A second antibiotic (usually a glycopeptide) was added in 20% and 21% of the cases, respectively. Overall, the response to therapy, with or without an additional antibiotic, was 95% (cefepime) and 90% (imipenem-cilastatin). Survival was similar in both groups (95% and 98%, respectively). Cefepime treatment was better tolerated, with 9% of the patients experiencing related intercurrent events compared with 19% in the imipenem-cilastatin group (P = 0.003). Nausea/vomiting was significantly more frequent in the imipenem-cilastatin group (15%) than in the cefepime group (5%; P = 0.001). Cefepime monotherapy was as effective as, and better tolerated than, imipenem-cilastatin in the empirical treatment of fever during short duration neutropenia.