Thirty-one moderately or severely ill hospitalized patients with proved (25 patients) or suspected (six) bacterial infections were randomly allocated to receive imipenem/cilastatin (16) or cefotaxime (15).The median age, sex, duration of therapy, underlying disease, and types of infection were similar in both groups. Nineteen patients with pneumonia, eight with soft tissue infection, and four with acute pyelonephritis were included. The pathogens isolated included Escherichia coli (six), Streptococcus pneumoniae (five), Streptococcus pyogenes (five), Haemophilus species (four), Proteus species (three), Staphylococcus aureus (three), and Serratia marcescens (two). In the imipenem/cilastatin group, 13 patients were cured of their infections and three showed improvement. In the cefotaxime group, nine were cured, three showed improvement, and three showed no improvement. Nine patients treated with imipenem/cilastatin developed phlebitis, as compared with eight treated with cefotaxime. One patient treated with cefotaxime developed diarrhea. During therapy, potential pathogens were isolated from four patients in the imipenem/cilastatin group (Candida species [two] and Pseudomonas maltophilia [two]), as compared with eight in the cefotaxime group (enterococci [two], Pseudomonas aeruginosa [two], Candida species [two], Acinetobacter anitratus [one], and Pseudomonas fluorescens [one]). There were no recognized superinfections.