The potential for bacterial contamination during transvaginal recovery of endometrial samples in patients with postpartum endomyometritis spawned an investigation into various protective measures and techniques. The confirmation and determination of the extent of contamination was attempted in 120 postpartum women with the diagnosis of postpartum endometritis. Each woman underwent endometrial sampling for bacterial culturing utilizing two devices--an unprotected, cotton-tipped swab and a protective, double-lumen catheter with a brush. Although a statistically greater number of bacterial species were recovered with the swab than with the brush (2.4 vs. 1.6, respectively [P less than .001]), they were not representative of species commonly thought of as causing cervicovaginal contamination. Individual species found with greater frequency with the swab were Streptococcus faecalis (P = .023), Escherichia coli (P = .007), Proteus mirabilis (P = .025), Bacteroides bivius (P less than .001 and Bacteroides melaninogenicus (P = .02). Thus, with proper cervical preparation and an extended-reach, transvaginal, cotton-tipped swab, true fundal specimens were obtained that provided a representative and cost-efficient sample of the infected endometrial cavity.