To assess the efficacy of single-agent therapy relative to standard combination antibiotic therapy for the initial management of fever and neutropenia in cancer patients, we conducted a randomized trial comparing ceftazidime alone with a combination of cephalothin, gentamicin, and carbenicillin. Of 550 evaluable episodes of fever and neutropenia, 282 were treated with ceftazidime alone and 268 with the combination. All episodes were evaluated for responses at 72 hours after the start of treatment and at resolution of the neutropenia. Of the patients with unexplained fever who were given ceftazidime alone, 99 percent were alive at 72 hours and 98 percent were alive when the neutropenia resolved, as compared with 100 percent and 98 percent, respectively, of those given combination therapy. Of the patients with documented infection who were given ceftazidime alone, 98 percent were alive at 72 hours and 89 percent when the neutropenia resolved, as compared with 98 percent and 91 percent, respectively, of those given combination therapy. The majority of episodes of documented infection in both treatment groups necessitated additional antimicrobial treatment or other modifications of the initial regimen, as compared with only 22 percent of the episodes of unexplained fever. We conclude that initial single-agent therapy with certain beta-lactam antibiotics is a safe alternative to standard combination antibiotic therapy, although patients with documented infection or protracted neutropenia are likely to require additional or modified treatment.