Abstract Study question Can extracellular vesicles serve as biomarkers of endometriosis, particularly for stages I and II, which are difficult to detect by ultrasound? Summary answer Our findings suggest the feasibility of defining a protein signature of endometriosis based on EV cargoes. What is known already Endometriosis is characterised by intraperitoneal lesions of endometrial tissue. Although ultrasound detection of severe endometriosis is possible, the diagnosis of peritoneal lesions currently relies on invasive laparoscopy, which leads to a delay of 8-10 years. We have previously shown that peritoneal fluid (PF) of women with endometriosis contains endometriosis specific EVs. Traced in peripheral blood (PB), EVs could serve as biomarkers of endometriosis, particularly for stages I and II, which are difficult to detect by ultrasound. Study design, size, duration Women 18-45 years of age undergoing laparoscopic surgery for endometriosis or unrelated conditions were invited to participate in our study (HREC 20-159A/HREC 20-882A). Following informed consent, we collected PF and PB samples from n = 101 participants (n = 47 controls, n = 54 cases). Exclusion criteria were pregnancy, malignancy, and menopause. Participants/materials, setting, methods We processed PF and PB samples by differential ultracentrifugation and validated the presence of EVs by nanoparticle tracking analysis (NTA), Western blotting and transmission electron microscopy (TEM). EVs were analysed by untargeted, label-based quantitative proteomics using Tandem Mass Tags (TMT). Data analysis was performed using Proteome Discoverer v2.4 (Thermo Fisher) followed by statistical analysis using R. Candidate biomarker proteins were tested by traditional and automated Western blotting (WES, ProteinSimple). Main results and the role of chance We ascertained the identity of EVs by TEM and NTA (mode size 121.8 ± 18.0 nm (blood, n = 5) and 155.9 ± 37.2 (PF, n = 6)), and signals for syntenin and ALIX in immunoblots as well as expression of CD9, CD63 and CD81 in capture bead-flow cytometry. Proteomic analysis identified 9145 protein groups across all sample groups, with 602 significantly regulated between comparisons (adjusted P-value <0.05). In PF, 533 proteins changed significantly in abundance (245 up/288 down), while in blood, only four proteins changed in abundance. We tested the presence/absence of two of these as candidate markers on PB EV protein preparations from n = 32 samples to date; our best marker showed a sensitivity of 67% and a specificity of 90%; upon exclusion of severe endometriosis cases, sensitivity improved to 80%. Limitations, reasons for caution Because of the changes in vesicle composition and amounts with the menstrual cycle, we have to group the samples by menstrual cycle phase and disease severity. This leaves relatively small numbers per group per condition. Another limitation concerns the relatively small amount of sample liquid and thus, EVs available for analysis. Wider implications of the findings Our findings suggest the feasibility of defining a protein signature of endometriosis based on EV cargoes. A non-invasive clinical test would allow for earlier diagnosis and thus treatment of endometriosis. It would also allow patients to initiate fertility treatment early. Trial registration number not applicable