The accuracy of high-risk human papillomavirus testing with the Xpert HPV assay on vaginal self-samples was compared with clinician-taken samples within the VALidation of HUman papillomavirus assays and collection DEvices for Self-samples and urine samples (VALHUDES) framework. Five-hundred and twenty-three women were recruited in five Belgian colposcopy clinics, of whom 483 (median age, 40 years; interquartile range, 31 to 49 years) were included in the main analysis (226 collected with Evalyn Brush and 257 collected with Qvintip). Cervical samples were collected with Cervex-Brush. Colposcopy and histology outcomes were considered as the reference standard. The Xpert HPV assay had similar accuracy for cervical intraepithelial neoplasia ≥2 on self-collected versus clinician-collected samples [relative sensitivity, 0.96 (95% CI, 0.91-1.02); and relative specificity, 0.96 (95% CI, 0.89-1.04)]. The relative accuracy slightly differed by vaginal collection device [sensitivity ratios of 0.98 (95% CI, 0.90-1.06) and 0.94 (95% CI, 0.87-1.02) for Evalyn and Qvintip, respectively; specificity ratios of 1.06 (95% CI, 0.95-1.19) and 0.88 (95% CI, 0.80-0.98) for Evalyn and Qvintip, respectively]. No difference in cycle threshold values was observed between vaginal and cervical samples. In conclusion, the sensitivity of Xpert HPV assay for cervical intraepithelial neoplasia ≥2 on vaginal self-samples was similar to that of cervical specimens. The clinical specificity was lower than on clinician-collected samples when self-samples were taken with Qvintip.