A large number of very early spontaneous preterm births are ascribed to bacterial infection of the upper genital tract or chorioamnionitis, without symptoms in either instance. Many believe that ascending infection by bacteria initially present in the cervix and vagina leads to microbial colonization of the chorioamnion and amniotic fluid. This study was an attempt to determine whether antibiotic treatment during the interpregnancy interval in women having a previous preterm birth before 34 weeks gestational age might lower the risk of spontaneous preterm birth in a subsequent pregnancy. A total of 241 women with a history of spontaneous birth before 34 weeks gestational age were randomized, in a double-masked trial, to receive either 2 1-g doses of azithromycin orally 4 days apart plus 750 mg daily of sustained-release metronidazole for 1 week or identical-appearing placebos. The regimen was repeated at 4-month intervals until a subsequent pregnancy began, as was the case for 124 women, 59 of whom received antibiotics. The antibiotic-treated and control women were similar in most respects, but the actively treated women were significantly more compliant in keeping study appointments and taking their assigned medications. Approximately 80% of women in both groups received a course of treatment within 6 months before the subsequent conception. Positive endometrial cultures at the time of randomization were comparably frequent in the antibiotic and placebo groups: 86% and 91%, respectively. Median numbers of individual microbial species isolated from the endometrium also did not differ significantly. Plasma cell endometritis was found in 31% of the antibiotic group and 47% of placebo recipients. Repeat endometrial cultures taken 2 weeks later showed a lower number of individual bacterial species in 49% of women given antibiotics and 31% of the control group. There was no apparent difference in the resolution of endometritis. The risk of a subsequent preterm birth did not differ significantly between the 2 groups, and the same was the case for miscarriage before 15 weeks gestation. Although not statistically significant, average gestational age at delivery was 2.4 weeks shorter in the antibiotic group, and average birth weight was 418 g lower than in the placebo group.