The vagina is not sterile and is not supposed to be. It hosts a dense bacterial community — hundreds of millions of organisms per millilitre of fluid — and the composition of that community turns out to matter a great deal for infection risk, for pregnancy outcomes, and for a set of symptoms women are commonly told are just bad luck.
This is a relatively young field. The sequencing technology that made it possible to describe these communities properly arrived in the mid-2000s, and much of what is now standard was unknown twenty years ago. It is also a field with more confident marketing than confident evidence, so this guide tries to be clear about which parts are well established and which are not.
It is mostly one genus, and that is unusual
Most bacterial communities in the human body are diverse. The gut microbiome of a healthy adult contains hundreds of species, and that diversity is a marker of health — low gut diversity is associated with disease.
The healthy vaginal microbiome is the exception. In many women it is dominated by a single genus, Lactobacillus, often by a single species, which can account for 90% or more of the community. Here, low diversity is the healthy state and high diversity is associated with problems.
This inversion catches people out, including people who have read about the gut microbiome and reasonably assume the principle transfers. It does not.
Community state types
In 2011, a study of nearly 400 women published in the Proceedings of the National Academy of Sciences grouped vaginal microbiomes into five recurring patterns, now called community state types, or CSTs. The framework is still in use.
- CST-I — dominated by Lactobacillus crispatus
- CST-II — dominated by Lactobacillus gasseri
- CST-III — dominated by Lactobacillus iners
- CST-IV — diverse, low in lactobacilli, with anaerobes such as Gardnerella, Prevotella and Atopobium
- CST-V — dominated by Lactobacillus jensenii
Two things about this list are worth absorbing.
First, the four Lactobacillus-dominated types are not equivalent. CST-I, dominated by L. crispatus, is the most stable and most consistently associated with low infection risk. CST-III, dominated by L. iners, is considerably less stable — L. iners is a lactobacillus, but it behaves differently from the others and transitions to CST-IV more readily.
Second, CST-IV is not an infection. It is a community composition. Many women with CST-IV have no symptoms at all. It is, however, the state associated with bacterial vaginosis, with increased susceptibility to sexually transmitted infection, and with increased risk of urinary tract infection.
CST distribution also varies by population and ethnicity, which is relevant because most of the foundational studies were conducted in North America and Europe. Indian data is comparatively thin — one of several reasons this field needs more work in Indian populations.
How the defence actually works
The protective mechanism is better understood than most people assume, and the details matter because they rule out some popular explanations.
Lactic acid
This is the main event. Lactobacillus species ferment sugars and produce lactic acid, which keeps vaginal pH in the range of roughly 3.8–4.5 — genuinely acidic, comparable to tomato juice.
A 2011 study in BMC Infectious Diseases tested this directly. Working under anaerobic conditions designed to approximate the real vaginal environment, the researchers exposed seventeen species of bacterial-vaginosis-associated bacteria and four species of vaginal lactobacilli to lactic acid. At physiological concentrations (55–111 mM) and pH 4.5, lactic acid inactivated all seventeen of the BV-associated species, while leaving the vaginal lactobacilli unharmed.
That is a striking degree of selectivity. The acid produced by the resident community suppresses its competitors and spares its producers.
What about hydrogen peroxide?
For years, hydrogen peroxide production was cited as a key protective mechanism, and it still appears in a great deal of marketing material.
The same 2011 study tested this too, and the result went the other way. At physiological concentrations (under 100 μM), hydrogen peroxide failed to inactivate any of the BV-associated species tested — even with human myeloperoxidase added to boost its activity. At high concentrations it did kill bacteria, but it killed the vaginal lactobacilli more readily than the BV-associated organisms.
A 2018 analysis in Microbiome reviewed the question and argued that the in vivo role of hydrogen peroxide as a vaginal antimicrobial is implausible, given the anaerobic conditions and the antioxidants present in vaginal fluid and semen.
The honest position is that lactic acid is the well-supported mechanism and hydrogen peroxide is not. If you see a product claiming benefit on the basis of hydrogen-peroxide-producing strains, that claim is weaker than it sounds.
Physical and immune contributions
Lactobacilli occupy the epithelial surface and compete for attachment sites and nutrients, which makes colonisation by incoming organisms harder — straightforward ecological competition. Some strains also produce bacteriocins, narrow-spectrum antibacterial peptides. And a Lactobacillus-dominant community is associated with a quieter, less inflammatory immune state at the vaginal surface, which keeps the epithelial barrier intact.
What feeds it: glycogen and oestrogen
The community needs a carbon source, and the main one is glycogen.
Vaginal epithelial cells store glycogen, and they do so under oestrogen control. This explains a pattern that otherwise looks arbitrary: the Lactobacillus-dominant state is characteristic of the reproductive years, is largely absent before puberty, and declines after menopause. It tracks oestrogen, because oestrogen determines the food supply.
A 2019 comparative genomics study in Microbiome examined how L. crispatus accesses this glycogen. The researchers identified a pullulanase type I gene encoding a cell-surface-anchored enzyme capable of breaking down glycogen outside the cell, and found that variation in this gene across strains correlated with their ability to grow on glycogen — strains with deletions in the signal peptide region showed reduced or absent glycogen utilisation.
That is a genomic correlation across strains, which is worth stating precisely. It is a plausible mechanism for why some L. crispatus strains outcompete others, and it has not been demonstrated that pullulanase activity is required for dominance in a living woman. The finding is suggestive and useful; it is not settled.
The oestrogen link also explains why genitourinary syndrome of menopause exists as a clinical entity. Falling oestrogen means less epithelial glycogen, less substrate for lactobacilli, a declining Lactobacillus population, rising pH, and a documented increase in urinary tract infection incidence.
Why this connects to urinary infections
The vagina and the urethral opening are a few centimetres apart, and vaginal colonisation is a recognised step in the development of urinary infection. Bacterial vaginosis is associated with elevated rates of recurrent UTI.
There is also a more specific mechanism, demonstrated in mice. A 2017 study in PLoS Pathogens established dormant E. coli reservoirs inside mouse bladder cells — these reservoirs are a known phenomenon and are thought to explain why recurrent infections are often caused by the same strain as the original. The researchers then exposed the bladders briefly to Gardnerella vaginalis, an organism characteristic of CST-IV.
G. vaginalis did not establish an infection itself. It was cleared quickly. But its brief presence triggered the dormant E. coli to emerge and cause active infection, damaged the bladder lining, and produced kidney injury that persisted after the Gardnerella was gone. Exposure to L. crispatus, or to heat-killed Gardnerella, did nothing.
The authors called this “covert pathogenesis”: an organism causing disease without being the infecting agent. It is a strong candidate explanation for why women with disrupted vaginal communities get recurrent urinary infections.
Two caveats, stated plainly. This is a mouse model, and it has not been demonstrated in women. And it is a mechanistic explanation for an epidemiological association, not proof that this mechanism accounts for the association.
What actually disrupts the community
Evidence quality varies considerably across this list.
Well supported: antibiotics, particularly broad-spectrum courses, which deplete lactobacilli along with the target organism; falling oestrogen at menopause and postpartum; douching, which physically removes the community and raises pH.
Reasonably supported: menstruation, which transiently raises pH; semen, which is alkaline; some hormonal contraceptives.
Emerging, and relevant in India: water and sanitation exposure. A 2024 prospective cohort in BMC Infectious Diseases followed 436 adolescent girls in western Kenya over 18 months with 16S rRNA sequencing and found that odds of CST-I were higher among those using piped water compared with pond water, and lower among those using traditional pit latrines compared with flush toilets. Coliform abundance was higher with stream water and non-flush latrines. Increased rainfall was associated with lower odds of CST-I.
That is the most direct evidence currently available that household water source and sanitation type track with vaginal microbiome composition. It is from Kenya rather than India, and it is observational, but it is directly relevant to a country where a large proportion of the population lacks safely managed water.
What the evidence does not support
Being clear about this matters, because the gap between the science and the marketing in this category is wide.
Restoring a disrupted community is harder than it sounds. The best-known trial of intravaginal L. crispatus for preventing recurrent UTI found recurrence in 15% of the treatment group versus 27% on placebo — but the 95% confidence interval ran from 0.2 to 1.2, which crosses 1. The result was not statistically significant, despite being widely cited as though it were. A Cochrane review of probiotics for UTI prevention found no significant benefit (RR 0.82, 95% CI 0.60–1.12).
The picture is better for bacterial vaginosis. The same strain significantly reduced BV recurrence in a 2020 New England Journal of Medicine trial — 30% versus 45% on placebo, RR 0.66, 95% CI 0.44–0.87.
So: restoring the community is achievable and helps with BV. Translating that into fewer urinary infections is not yet demonstrated. Anyone telling you otherwise is ahead of the data.
This article is general information and is not medical advice. If you have symptoms, see a doctor.
Sources
- Ravel J, et al. Proc Natl Acad Sci USA. 2011;108(Suppl 1):4680-4687. doi:10.1073/pnas.1002611107
- O’Hanlon DE, Moench TR, Cone RA. BMC Infect Dis. 2011;11:200. doi:10.1186/1471-2334-11-200
- Tachedjian G, O’Hanlon DE, Ravel J. Microbiome. 2018;6:29. doi:10.1186/s40168-018-0418-3
- van der Veer C, et al. Microbiome. 2019;7:49. doi:10.1186/s40168-019-0667-9
- Gilbert NM, O’Brien VP, Lewis AL. PLoS Pathog. 2017;13(3):e1006238. doi:10.1371/journal.ppat.1006238
- Czapar AE, et al. BMC Infect Dis. 2024;24(1):1419. doi:10.1186/s12879-024-10313-3
- Stapleton AE, et al. Clin Infect Dis. 2011;52(10):1212-1217. doi:10.1093/cid/cir183
- Cohen CR, et al. N Engl J Med. 2020;382(20):1906-1915. doi:10.1056/NEJMoa1915254
- Schwenger EM, Tejani AM, Loewen PS. Cochrane Database Syst Rev. 2015;(12):CD008772
- Portman D, Gass M. J Sex Med. 2014;11(12):2865-2872
