The purpose of this investigation was to evaluate the efficacy of amoxicillin for treatment of bacterial vaginosis during pregnancy. The diagnosis of bacterial vaginosis was established by clinical examination and microscopic examination of a Gram stain and saline preparation of vaginal secretions. In a double-blind, randomized manner, 108 patients at 15-25 weeks' gestation were assigned to treatment with oral amoxicillin, 500 mg three times daily for 14 days, or placebo. Patients were evaluated 2 weeks after treatment, at 34-36 weeks' gestation, and at delivery. There were no significant differences between the two groups with respect to any clinical or microbiologic measure of treatment outcome. There were also no significant differences in the frequency of obstetric complications. We conclude that amoxicillin is not effective therapy for bacterial vaginosis in pregnant women.