In a randomized multicentre study ciprofloxacin combined with azlocillin was compared with gentamicin and azlocillin for the treatment of febrile episodes in neutropenic patients. In 147 evaluable episodes in 108 patients, 80 patients received ciprofloxacin/azlocillin and 67 received gentamicin/azlocillin. The two treatment groups were comparable in terms of age, underlying diagnosis, and duration of neutropenia. Microbiologically documented infections were the cause of fever in 34 (42.5%) and 29 (43.3%) episodes in the ciprofloxacin/azlocillin and gentamicin/azlocillin groups respectively. At the end of therapy, 46 patients (57.5%) receiving ciprofloxacin/azlocillin showed complete resolution compared with 30 (44.7%) for the gentamicin/azlocillin group (P = 0.14). The clinical response rate for microbiologically documented episodes was 58.8% and 48.3% respectively (P = 0.45). Among the microbiologically documented infections with follow-up cultures available, 24 (92.3%) of 26 isolates from patients receiving ciprofloxacin/azlocillin were eradicated, in comparison with 19 (86.4%) of 22 in the gentamicin/azlocillin group (P = 0.65). There were five superinfections, all in the gentamicin/azlocillin group. Significant resistance to the study drugs was not seen. Of all evaluable patients, including those subsequently withdrawn because of early modification of therapy, there were 12 deaths within the study period; six (6.8%) of these occurred in 88 patients randomized to the ciprofloxacin/azlocillin group, compared with two of 80 (2.5%) in the gentamicin/azlocillin group. Both treatments were generally well-tolerated; one patient in the ciprofloxacin/azlocillin group developed convulsions, probably related to ciprofloxacin. The combination of ciprofloxacin and azlocillin is as effective as gentamicin plus azlocillin and offers a useful alternative for the empirical treatment of febrile neutropenic patients.