To describe the microbiologic etiology of post-cesarean endometritis developing after perioperative cephalosporin prophylaxis, endometrial samples were obtained from 27 women with a triple-lumen catheter. The women were assigned in a double-blind, randomized fashion to receive either ticarcillin/clavulanate, 3.1 g, or cefoxitin, 2 g, administered every six hours, until the clinical signs of infection resolved. A total of 149 microorganisms (84 facultative and 65 obligate anaerobes) were recovered from 26 women, for a mean of 5.5 isolates per specimen. One endometrial specimen was sterile. Bacteroides and Peptostreptococcus species were the most frequent isolates, followed by Gardnerella vaginalis, Enterococcus, facultative gram-negative rods and Mycoplasma hominis. Each of the isolates was tested for beta-lactamase activity. At least one beta-lactamase-producing isolate was recovered from 56% of the specimens. Susceptibility testing of endometrial isolates demonstrated that 96% of 118 potential pathogens (Gardnerella, Bacteroides, Peptostreptococcus, enterococci and streptococci) were susceptible to ticarcillin/clavulanate. By comparison, 86% of these isolates were susceptible to cefoxitin in vitro. Women who were treated with ticarcillin/clavulanate were less likely to have a temperature greater than 38 degrees C for two or more days (8% vs. 57%, P = .01). Also, there was a trend toward a decreased duration of uterine tenderness in the ticarcillin/clavulanate group, but it did not attain statistical significance (60% vs. 86%, P = .4). However, the overall clinical success rate with these single-agent treatments was not different for the two groups (77% vs. 79%, P = 1.0).(ABSTRACT TRUNCATED AT 250 WORDS)