The introduction of self‐sampling in cervical cancer screening has raised the importance of HPV test validation on self‐collected samples. This study aimed to evaluate the clinical accuracy of the OncoPredict HPV Screening (SCR) assay on self‐collected vaginal and first‐void urine (FVU) samples, relative to cervical specimens, using the VALHUDES Framework. FVU and vaginal self‐samples followed by a clinician‐collected cervical brushing were collected from 500 women referred to colposcopy and tested using OncoPredict HPV SCR assay. The assay demonstrated clinical sensitivity to detect cervical intraepithelial neoplasia grade 2 or worse (≥ CIN2) similar to cervical samples in FVU (ratio: 0.95, [95% CI: 0.88–1.02]) and vaginal self‐samples (ratio: 0.96 [95% CI: 0.90–1.02]). The clinical specificity for < CIN2 was lower in vaginal (ratio: 0.90 [95% CI: 0.84–0.96]) but not in FVU samples (ratio: 1.03 [95% CI: 0.96–1.12) when compared to cervical samples. However, the relative specificity improved following cut‐off optimization (ratio: 0.94, 95% CI: [0.88–1.01]). Moderate to excellent agreement in HPV detection between self‐collected and cervical samples was demonstrated (Kappa values: 0.53–1.00). To conclude, OncoPredict HPV SCR assay demonstrated similar accuracy on FVU and cervical samples. On vaginal compared to cervical samples sensitivity was similar with a lower specificity, which improved with cut‐off optimization.