The validity of testing for high-risk types of human papillomavirus (HPV) in cervical cancer prevention programs is undetermined. We compared the performance on primary screening of HPV DNA testing, cytology and colposcopy in detecting cervical intra-epithelial neoplasia (CIN) grade 2 or 3 or cancer. A cohort of 4,761 women, median age 35 years, was screened by routine cytology, routine colposcopy and testing for high-risk HPV by a PCR-based method. Within an 8-month period, women with abnormal findings on cytology or screening colposcopy or in whom high-risk HPV types were detected were referred for colposcopy and biopsy. Women negative on all initial screening tests were followed by a second screening examination. To correct for work-up bias, the true prevalence of CIN 2 or 3 or cancer was estimated by projection from histologically verified subgroups. Cervical biopsies were taken in 364 women (7.6%), of whom 114 (2.4%) showed CIN 2 (n = 34) or CIN 3 (n = 71) or cancer (n = 9). High-risk HPV testing achieved bias-corrected performance measures of 89.4% sensitivity, 93.9% specificity, 35.8% positive predictive value and 99.6% negative predictive value. Bias-corrected rates of true- and false-positives by high-risk HPV testing compared to cytology (colposcopy) were about 4.5 (6.7) and 19.1 (7.4) times higher, respectively. The quality of routine cytology was controlled by computer-assisted review, and the observed number of true-positives more than doubled after adding automated review results. In middle-aged women, testing for high-risk HPV types, particularly when negative, may be used to increase the screening interval in programs for secondary prevention of cervical cancer. Int. J. Cancer 89:529–534, 2000. © 2000 Wiley-Liss, Inc.