Introduction: The clinical course of high-grade cervical intraepithelial neoplasia (CIN2/3) is characterised by a high spontaneous regression rate. Histological assessment is unable to differentiate between CIN2/3 lesions likely to regress and those likely to persist or progress. Most CIN2/3 lesions are treated by surgical excision, leading to overtreatment of a substantial proportion. In this prospective study, we evaluate the value of DNA methylation of host cell genes, which has shown to be particularly sensitive for the detection of advanced CIN2/3 and cervical cancer, in the prediction of regression or non-regression of CIN2/3 lesions. Methods and analysis: This is a multicentre observational longitudinal study with 24-month follow-up. Women referred for colposcopy with an abnormal cervical scrape, who have been diagnosed with CIN2/3 and a small cervical lesion (≤50% of cervix) will be asked to participate. Participants will be monitored by 6-monthly cytological and colposcopic examination. In case of clinical progression, participants will receive treatment and exit the study protocol. At baseline and during follow-up, self-sampled cervicovaginal brushes and cervical scrapes will be collected for high-risk human papillomavirus (HPV) testing and FAM19A4/miR124-2 methylation analysis. A colposcopy-directed biopsy will be taken from all participants at the last follow-up visit. The primary study endpoint is regression or non-regression at the end of the study based on the histological diagnosis. Regression is defined as CIN1 or less. Non-regression is defined as CIN2 or worse. The secondary study endpoint is defined as HPV clearance (double-negative HPV test at two consecutive time-points). The association between methylation status and regression probability will be evaluated by means of χ 2 testing. Ethics and dissemination: Ethics approval was obtained in all participating clinics. Results: of the main study will be submitted for publication in a peer-reviewed journal.