Diverse and non-Lactobacillus-dominated vaginal microbial communities are associated with adverse health outcomes such as preterm birth and acquisition of sexually transmitted infections. Despite the importance of recognizing and understanding the key risk-associated features of these communities, their heterogeneous structure and properties remain ill-defined. Clustering approaches have been commonly used to characterize vaginal communities, but they lack sensitivity and robustness in resolving community substructures and revealing transitions between potential sub-communities. We used a more highly resolved approach based on mixed membership topic models with multi-domain longitudinal data from cohorts of pregnant and non-pregnant subjects to identify several non-Lactobacillus-dominated sub-communities common to women regardless of reproductive status. These sub-communities correlated with clusters of metabolites. In non-pregnant subjects, we identified a few sub-communities that were more common during menses but did not predict an increased likelihood of non-Lactobacillus-dominated communities during the rest of the menstrual cycle. The menstrual cycle was a strong driver of transitions between sub-communities and was correlated with changes in levels of cytokines, for example, elevated TNF-α concentrations at the time of ovulation, and metabolites, for example, elevated kynurenine concentrations during menses. In pregnant women, some metabolite clusters were predictive of changes in vaginal microbiota structure. Overall, our results show that the vaginal community substructure is shaped by the menstrual cycle and that specific sets of metabolites are associated with community instability during pregnancy.