OBJECTIVES: The aim of this study was to examine the accuracy of the presence of high-risk human papillomavirus (HR-HPV) DNA (HR-HPV DNA test) postconisation as prediction of recurrent or residual cervical intraepithelial neoplasia (CIN) after treatment of high-grade cervical intraepithelial lesions (CIN2+) in a prospective study and to compare this with follow-up cytology and the marginal status of the excised tissue. DESIGN: Prospective follow-up study. SETTING: Unselected women presenting at colposcopy clinic of University Hospital Gasthuisberg, Leuven. POPULATION: Seventy-two women treated with conisation for CIN2 or CIN3. METHODS: Women were followed by HR-HPV DNA test (Hybrid Capture II test of Digene) every 3 to 6 months. The same vial was used for cytology and the HR-HPV DNA test (SurePath). All women were further followed by colposcopy and cytology for 24 months at 6-month intervals. The outcome of the study was presence of >CIN2, proven with colposcopy-directed biopsy occurring within 24 months after treatment. HR-HPV status was correlated with recurrent or residual CIN2+. MAIN OUTCOME MEASURES: Sensitivity, specificity, predictive values and diagnostic odds ratios to predict treatment failure or cure were computed for HR-HPV testing, marginal status and follow-up cytology. HR-HPV status was also correlated with section margins postconisation and with the first cervical smear. RESULTS: In 6 of the 72 treated women (8%), residual or recurrent CIN occurred. Women with recurrence were significantly older than women without a recurrence (51.5 +/- 9.6 versus 39.8 +/- 12.2 years, P= 0.007). All six women with recurrence were HR-HPV positive, four had a positive follow-up smear (>or=atypical squamous cells of uncertain significance = ASCUS+) and only two had involved section margins. Among the 66 cured women, 15 were HR-HPV positive, 6 had an abnormal smear and 12 had positive section margins. Sensitivity of cytology, positive section margins and HR-HPV DNA positivity was 66.7, 33.3 and 100% to predict treatment failure. Specificity of the three tests was, respectively, 90.9, 81.8 and 77.3%. Women with HR-HPV DNA at 3 to 6 months showed recurrent or residual CIN in 15% (2/13) if they had normal follow-up Pap smears and in 50% (4/8) if they had abnormal Pap smears. Margin status was not statistically significantly associated with human papillomavirus status. CONCLUSION: Persistence or clearance of HR-HPV DNA is an early valid prognostic marker of failure or cure after treatment for CIN2+ and is more accurate than cytology or section margin status at the time of conisation. The absence of HR-HPV DNA has a 100% negative predictive value. Higher age at conisation may be a previously unrecognised risk factor for recurrence.