BACKGROUND: Bacterial vaginosis (BV) and Trichomonas vaginalis (TV) have been estimated to affect one-quarter to one-third of sexually active women worldwide, and are often found concurrently. Few studies have examined this relationship longitudinally to better understand the direction and temporality of this association. METHODS: Between 2005 and 2006, a cohort of 853 young, sexually active women was followed in Mysore, India; participants were interviewed and tested for BV and TV at baseline, and at 3- and 6-month visit. Generalized estimating equations were used to estimate how changes in vaginal flora between consecutive visits-as defined by Nugent diagnostic criteria for BV-were related to the risk of TV infection at the latter visit, adjusted for sociodemographic and behavioral covariates. Treatment was offered to women with TV and/or symptomatic BV. RESULTS: After adjustment for covariates, participants with abnormal vaginal flora at 2 consecutive visits had 9 times higher risk of TV (95% CI: 4.1, 20.0) at the latter visit, relative to those with persistently normal flora. An increased risk of TV was also observed for participants whose flora status changed from normal to abnormal (adjusted risk ratio: 7.11, 95% CI: 2.8, 18.2) and from abnormal to normal (adjusted risk ratio: 4.50, 95% CI: 1.7, 11.8). CONCLUSIONS: Women experiencing abnormal flora during a 3-month span appear to have significantly increased risk of acquiring TV infection. Women of reproductive age in low-resource settings found to have abnormal vaginal flora should be assessed for TV.