In The Netherlands and most other European countries, women with two serial cervical smears with borderline or mild dyskaryosis (BMD) within 6 months are referred for colposcopy-directed biopsies. Only about 10% of these women have high-grade cervical intraepithelial neoplasia (CIN). This study therefore investigated whether human papillomavirus (HPV) testing could identify which women with smears read as BMD are most likely to have high-grade CIN, either at referral or during follow-up and the relationship was determined between clearance of high-risk HPV and regression of abnormal cytology. Women with smears read as BMD (n=278) were referred to the gynaecologist for colposcopy. They were subdivided into two groups; group A comprised women with a single smear (n=172) and group B women with two sequential smears (n=106) read as BMD before referral. High-risk HPV detection with Hybrid Capture II (HC II) was performed on a cervical scrape taken at the first visit before colposcopy (i.e. baseline smear) and during follow-up. Biopsies were taken when lesions suspected for CIN were seen at colposcopy. High-risk HPV DNA was present in the baseline smears of 126 (45.0%) women; 26 (20.6%) of them had histologically confirmed CIN 2/3 at the first visit and another 14 (11.1%) during follow-up. Only one of the 152 women (0.7%) with a negative high-risk HPV test had a CIN 2 lesion at the first visit and no CIN lesions were detected during follow-up of these women. After exclusion of women who were treated for prevalent high-grade CIN, the median follow-up times were 1.3 years (range 0.0–4.3 years) and 1.6 years (range 0.0–4.5 years) for women with HPV-negative and HPV-positive baseline smears, respectively. The sensitivity of a positive high-risk HPV test for CIN 2/3 at the first visit was 96.3%, the specificity 60.2%, the positive predictive value 20.6%, and the negative predictive value 99.3%. These values did not change markedly when stratified for group A or group B. Thus, a high-risk HPV positive test was strongly associated with the presence at the first visit and the development of CIN 2/3 lesions during follow-up. Moreover, regression of abnormal cytology in women with a positive high-risk HPV test at baseline was strongly associated with viral clearance and occurred 0.3 years (range −1.2 to 1.7 years) later than HPV clearance. This study establishes the value of a high-risk HPV positive test for women at risk of high-grade CIN, with virtually no risk for missing CIN 2/3. Addition of a test on high-risk HPV in women with BMD could prevent 55% of the referrals and/or repeat smears. Copyright © 2001 John Wiley & Sons, Ltd.