Objectives: To determine whether metronidazole treatment of bacterial vaginosis(BV) in pregnancy reduces the rate of spontaneous preterm birth (PTB). Study Design: Randomised, double-blind, placebo-controlled study of two successive monthly courses of oral metronidazole (400 mg bd for 2 days) in pregnant women at 24 weeks' gestation, with intermediate flora (heavy growth of G. vaginaljs] or bacterial vagjnosis (by Gram stain). Follow-up vaginal swabs were performed at 28, 32 and 36 weeks' gestation. Exclusions included women with age <17 years, multiple pregnancy, or recent antibiotics for vaginal discharge. Results 625 women fully completed the study protocol at one of the 4 participating centres. 17(5.5%) of 312 women in the placebo group gave birth preterm (<37 weeks), compared with 16(5.1%) of 313 in the metronidazole group. A direct smear indicative of BV was present in 341 women, however the PTB rate was similar in both the placebo (10/165) and metronidazole (10/176) groups. In contrast, treatment with metronidazole in women with obstetric/demographic risk factors for PTB, resulted in a reduction of the PTB rate. Among women with a previous preterm delivery (PPTD), 5(38.5%) of 13 given placebo gave birth preterm, compared with one(7.7%) of 13 given metronidazole. Other categories showing reductions in PTB rate were: "single" [10% placebo vs 6.2% metronidazole]. "age < 20 years" [ 11.4% vs 0%]. and "unemployed" [9.7% vs 3.9%], although these differences were not statistically significant. Conclusions: In women at low risk of PTB, treatment of BV in pregnancy appears to result in little or no reduction in the PTB rate, (although the power of this negative conclusion is limited due to sample size). In women with risk factors (especially PPTD), treatment of BV organisms during pregnancy may reduce the PTB rate, however further enrolments are necessary to demonstrate a statistically significant reduction.