A randomized comparison of ceftazidime versus clindamycin-tobramycin was performed for the treatment of obstetrical and gynecological infections. Entry criteria were an oral temperature of greater than or equal to 38 degrees C and a clinical diagnosis of endometritis, salpingitis, or pelvic cellulitis after hysterectomy. All patients with endometritis had cultures of intrauterine material obtained via a transcervical single-lumen catheter. The patients with pelvic cellulitis had material from the vaginal apex aspirated for culture, and all patients with salpingitis had a culdocentesis for culture of intraperitoneal material. Of 38 patients who received ceftazidime, 34 had endometritis after cesarean section, 3 had endometritis after abortion, and 1 had pelvic cellulitis. Of 39 patients who received clindamycin-tobramycin, 35 had endometritis after cesarean section, 3 had salpingitis, and 1 had pelvic cellulitis. The most common bacterial isolates were Lactobacillus sp., Bacteroides bivius, Escherichia coli, other gram-negative aerobic bacilli, group B streptococci, and other aerobic streptococci. Bacteremia occurred in 9.0% of the patients. Of the patients receiving clindamycin-tobramycin and ceftazidime, 34 (87.2%) and 34 (89.5%), respectively, responded to therapy. All the clinical failures occurred in patients with endometritis after cesarean section. Clinical failures had persistent fever despite 3 or more days of treatment. One of the patients receiving clindamycin-tobramycin developed an urticarial rash after her infection had resolved. No patient in either group developed diarrhea. In these small groups of patients, there were no significant differences in cure rate, side effects, or length of hospital stay.