The prevalence of human papillomavirus (HPV) genotypes in relation to age was investigated by the polymerase chain reaction (PCR) method in cytologically normal smears from 4 different groups of women. Group A consisted of young women from a district population, aged 15-34 years, using oral contraceptives and visiting general practitioners for a check-up (n = 156); group B were asymptomatic women, aged 35-55, in a district population participating in a triennial screening program for cervical cancer (n = 1555); group C and D consisted of women, seen at the gynecological outpatient department for a wide spectrum of gynecological complaints or for control of their hormonal contraception, aged 15-34 years (n = 2320), and aged 35-55 years (n = 1826) respectively. An HPV (all types) prevalence of 14.1%, 4.1%, 13.9% and 6.6% and an HPV 16/18 prevalence of 3.8%, 0.9%, 3.3% and 1.5% were found in groups A, B, C and D respectively. Statistically significant differences (p value < 0.001) in HPV prevalence were found between women aged 15-34 years and women aged 35-55 years in the district population and in the hospital population. No statistically significant differences in HPV 16/18 were observed after age-matching between women in corresponding age-classes of both populations. In a 5-year interval analysis a strong age-dependent relationship was demonstrated, with a maximum between 20 and 24 years. After the age of 35 a constant level of 1-2% HPV 16/18 was observed. These results indicate that genital HPV infections are age-dependent and suggest that HPV infections at young age can be transient. The implications of these findings in the context of cervical cancer screening are discussed.