The vaginal microbiome plays an important role in human health and species of vaginal bacteria have been associated with reproductive disease. Strain level variation is also thought to be important, but the diversity, structure and evolutionary history of vaginal strains is not as well characterized. We developed and validated an approach to measure strain variation from metagenomic data based on SNPs within the core-genomes for six species of vaginal bacteria: G. vaginalis, L. crispatus, L. iners, L. jensenii, L. gasseri, and A. vaginae. Despite inhabiting the same environment, strain diversity and structure varies across species. All species except L. iners are characterized by multiple distinct groups of strains. Even so, strain diversity is lower in the Lactobacillus species, consistent with a more recent colonization of the human vaginal microbiome. Both strain diversity and the frequency of multi-strain samples is related to species-level diversity of the microbiome in which they occur, suggesting similar ecological factors influencing diversity within the vaginal niche. We conclude that the structure of strain level variation provides both the motivation and means of testing whether strain level differences contribute to the function and health consequences of the vaginal microbiome.Data SummaryAll vaginal metagenomic sequence data generated for this project can be found on the Sequence Read Archive under BioProject PRJNA639592.