OBJECTIVE Our goal was to assess the relationship between bacterial vaginosis and sexual intercourse and the impact of both on preterm birth. STUDY DESIGN The presence of bacterial vaginosis was assessed in 790 healthy nulliparous women between 8 and 17 weeks' gestation, and they were then asked to record weekly the number of occasions of sexual intercourse from registration to term. The end point for each was the occurrence of preterm uterine contractions, preterm birth, or preterm rupture of the membranes. RESULTS A total of 407 women returned the specific follow-up charts on vaginal intercourse. Bacterial vaginosis was detected at registration in 93 of the 407 subjects (22.9%), but the presence or absence of bacterial vaginosis had no predictive value as regards the frequency of subsequent intercourse. Of the 39 women (9.6%) who experienced threatened preterm labor, 16 were delivered before 37 weeks of gestation. In addition, premature rupture of membranes occurred in 48 women (9 preterm and 39 term). The 95% confidence interval for the weekly frequency of intercourse was calculated for those women delivered at term; a larger (p < 0.05) proportion of women delivered preterm, with (6/9, 66.7%) or without (13/16, 81.3%) premature rupture of membranes, had frequencies less than the 95% confidence interval during the second trimester of pregnancy; the respective proportions were 75.0% (6/8) and 78.6% (11/14) during the third trimester. Furthermore, among those women who had stopped having sexual intercourse in the last trimester, a larger (p < 0.05) proportion was delivered preterm (57.1%), with or without premature rupture of membranes, than was delivered at term (13.6%). In addition, only three women (18.8%) delivered preterm, with (two women) or without (one woman) premature rupture of membranes, reported having intercourse within 1 week before the onset of complications. CONCLUSION For healthy nulliparous women, coitus during pregnancy is not related to bacterial vaginosis and does not predispose to preterm birth.