There is evidence linking bacterial vaginosis with preterm delivery, presently the leading cause of perinatal morbidity and death, as well as with neurodevelopmental delay. There also are indications that systemic antibiotic treatment lowers the risk of subsequent preterm birth. This randomized, placebo-controlled, double-blind trial was an attempt to learn whether antibiotic treatment of women with asymptomatic bacterial vaginosis (or intermediate abnormal flora) early in the second trimester lowers the risk of late miscarriage or preterm delivery. Among 6120 pregnant women screened at their first antenatal visit, at 12-22 weeks gestation, there were 494 who had either bacterial vaginosis (416) or abnormal vaginal flora (78) on Gram-stained slides of self-collected vaginal smears. These women were randomized to receive either 300 mg clindamycin or a placebo orally twice a day for 5 days. Miscarriage was defined as pregnancy loss at 13-24 weeks gestation, and preterm delivery as occurring at 24-37 weeks. Twenty-five women dropped out of the trial because of side effects. The clindamycin-treated patients had significantly fewer miscarriages and spontaneous preterm deliveries than did placebo recipients (5.3% vs. 15.7%). A majority of the 11 elective preterm deliveries were prompted by preeclampsia. The treatment and control groups did not differ significantly in gestational age, birth weight, or the proportion of newborn infants admitted to neonatal intensive-care units. Clindamycin had positive effects in women with and those without a history of spontaneous preterm birth or late pregnancy loss. The treatment effect appeared to be greater in those with more severe bacterial vaginosis. In a general obstetric population, treating asymptomatic bacterial vaginosis (or abnormal intermediate vaginal flora) in midpregnancy lowers the risk of late miscarriage or spontaneous preterm delivery. Further studies might show the best times to screen pregnant women and treat them.