Use of human papillomavirus (HPV) testing in cervical cancer prevention is increasing rapidly. A DNA test for 13 HPV types that can cause cervical cancer is approved in the United States for co-screening with cytology of women >or=30 years old and for triage of women of all ages with equivocal cytology. However, most infections with HPV are benign. We evaluated trade-offs between specificity and sensitivity for approximately 40 HPV types in predicting cervical intraepithelial neoplasia 3 and cancer in two prospective studies: a population-based screening study that followed 6196 women aged 30-94 years from Costa Rica for 7 years and a triage study that followed 3363 women aged 18-90 years with equivocal cytology in four U.S. centers for 2 years. For both screening and triage, testing for more than about 10 HPV types decreased specificity more than it increased sensitivity. The minimal increases in sensitivity and in negative predictive value achieved by adding HPV types to DNA tests must be weighed against the projected burden to thousands of women falsely labeled as being at high risk of cervical cancer.