Abstract Study question Is the composition of the endometrial microbiome comparable to that of the vagina and can vaginal sampling provide sufficient information to characterize the uterine microbiome? Summary answer The compositions of the endometrial and vaginal microbiome are largely comparable and therefore, in most cases, vaginal sampling is sufficient to characterize the uterine microbiome. What is known already A dysbiotic endometrial microbiome (i.e., a shift from a predominance of beneficial bacteria to less favourable species) has been reported in association with adverse pregnancy outcomes in assisted reproductive treatment (ART). If dysbiosis is detected, antibiotics or probiotics can be administered to restore an optimal microbiome. However, characterization of the endometrial microbiome is challenging, involving invasive sampling, and technical difficulties due to the low microbial mass in the endometrium. In contrast, vaginal sampling is considered minimally invasive and microbial analysis is technically less complex. It is important to clarify whether the microbiomes of the endometrium and the vagina are comparable. Study design, size, duration A prospective multi-centre study including 60 ART patients was conducted from June 2022 to August 2023. During the luteal phase (on average day-19), paired vaginal swabs and endometrial biopsies were collected sequentially and stored in a preservation medium for further analysis. Patients and physicians were also asked to complete a questionnaire to identify any lifestyle or medical factors that may have differentially affected the microbial composition of the vagina and endometrium. Participants/materials, setting, methods A carefully validated qPCR methodology was used to accurately assess the presence and relative abundance of the four major lactobacillus species associated with eubiosis (crispatus, gasseri, iners, and jensenii) and the 15 most common bacterial species implicated in dysbiosis. Additionally, the relative abundance of microbial and human DNA was examined. Of note, endometrial analysis required additional tissue lysis. Furthermore, due to the low microbial DNA content in endometrial samples, preamplification was necessary prior to qPCR. Main results and the role of chance The proportion of microbial versus human DNA was approximately 90 times higher in vaginal samples compared to endometrial samples (p < 0.0001). However, the microbial compositions of both the endometrium and the vagina showed remarkable similarities. Lactobacillus species were predominant at both sites, accounting for an average of 86% and 89% of the bacterial population in the vagina and endometrium, respectively (p = 0.6). Of the dysbiotic species, Gardnerella vaginalis was the most common, accounting for 8% (vagina) and 7% (endometrium) of the total bacterial population (p = 1). There was no difference in the average number of distinct microbial species between the two niches (1.8 vs. 2.1; p = 0.3). Direct comparison of paired samples highlighted the high similarity of microbial compositions between both sites and showed strong correlations of species abundances in individual patients (L.crispatus: r = 0.9, L.iners: r = 0.9, L.gasseri: r = 1, L.jensenii: r = 0.9, G.vaginalis: r = 0.7; p < 0.0001/all species). However, the microbial composition differed in four patients (6.6%), with three having a dysbiotic vaginal profile (≥50% dysbiotic species) but a eubiotic endometrial profile (≥50% lactobacillus species) and one showing the reverse pattern. The medical conditions of two patients, including chronic endometritis and use of antibiotics and probiotics in the preceding cycle, may account for the observed differences. Limitations, reasons for caution The species analysed were carefully selected to ensure a comprehensive analysis. Nevertheless, the potential presence of unevaluated species could impact the results. While the included patients displayed diversity in their reproductive histories, it is important to acknowledge that the observed results may still be influenced by underlying causes of infertility. Wider implications of the findings The close similarity between the microbiomes at the two sites investigated suggests that minimally invasive vaginal sampling may be sufficient to characterise the endometrial microbiome. The ease of vaginal sampling and processing is expected to improve patient recruitment in microbiome research and facilitate easier application of clinical microbiome tests. Trial registration number not applicable