Although a second age-related peak of human papillomavirus (HPV) infection is observed in many populations, it does not seem to have any impact on cervical screening policies. We examined the age-specific prevalence of HPV infection among 2,604 women enrolled for cervical screening and correlated the age at diagnosis of 2,491 cervical intraepithelial neoplasia Grade 2/3 (CIN2/3) lesions and 801 invasive cervical cancers (ICC). Two peaks of HPV infection were detected at 26-30 and 46-50 years, respectively. The first infection peak was followed by a CIN2/3 peak and an ICC peak at 5-15 and 15 years later, respectively. The second infection peak was followed by an ICC peak 20 years later, but strikingly no CIN2/3 peak was detected in between and thus eliminated an opportunity of treating the lesions at preinvasive stages. The most plausible explanation is that women at the expected second CIN2/3 peak (50-65 years) are not having Pap smears under the current opportunistic screening program. Furthermore, women of this age may have physiological retraction of the transformation zone, and CIN lesions may remain undetected if an adequate Pap smear sample is not obtained. To combat this problem, the screening program in Hong Kong needs to focus on women aged 50 years and older and a mop-up screening up to 75 years is necessary. Bimodal peaks of HPV infection and cervical cancer are seen in many countries and the analysis of populationspecific age distribution of CIN2/3 should be an integral exercise in evaluating the effectiveness of a screening program.Human papillomavirus (HPV) infection is a necessary cause of cervical cancer. 1 The infection is mainly transmitted by the sexual route and occurs most commonly in young sexually active individuals with peak prevalence in women younger than 25 years. 2,3 Interestingly, most studies have also observed a second less pronounced peak of infection among older women, mostly 55 years and older. 2 A few hypotheses have been proposed to explain this observation. First, hormonal changes associated with menopause might induce reactivation of latent HPV infection; however, this second peak has not been universally observed in postmenopausal women. Second, it could be attributed to changes in sexual behavior of women and their partners in middle age, but the occurrence of second peak did not correlate well with data on sexual behavior. 2,3 Finally, the second peak may be a reflection of population-specific cohort effects as it is not consistently observed across the world. 2,4 The underlying reason for this second minor peak is still obscure, and its potential importance has not been fully investigated. It has been shown that a3/a15 HPV types, including HPV61, 71, 72, 81, 83, 84 and 89, are more commonly found in cervical scrape samples collected from older women and these samples also contain more squamous cells. 5 Because a3/a15 HPV types are nononcogenic, the significance of the increased prevalence among older women is unclear. 6 Nonetheless, many studies have confirmed a second ...