Human papillomavirus has emerged as the leading infectious cause of cervical and other anogenital cancers. We have studied the relation between human papillomavirus infection and the subsequent risk of anal and perianal skin cancer. A case -cohort study within two large Nordic serum banks to which about 760 000 individuals had donated serum samples was performed. Subjects who developed anal and perianal skin cancer during follow up (median time of 10 years) were identified by registry linkage with the nationwide cancer registries in Finland and Norway. Twenty-eight cases and 1500 controls were analysed for the presence of IgG antibodies to HPV 16, 18, 33 or 73, and odds ratios of developing anal and perianal skin cancer were calculated. There was an increased risk of developing anal and perianal skin cancer among subjects seropositive for HPV 16 (OR=3.0; 95%CI=1.1 -8.2) and HPV 18 (OR=4.4; 95%CI=1.1 -17). The highest risks were seen for HPV 16 seropositive patients above the age of 45 years at serum sampling and for patients with a lag time of less than 10 years. This study provides prospective epidemiological evidence of an association between infection with HPV 16 and 18 and anal and perianal skin cancer. Anal epidermoid carcinoma is a rare tumour, but its incidence has increased over the past 30 years (Goldman et al, 1989;Frisch et al, 1993;Melbye et al, 1994). Sexual behaviour and smoking are risk factors (Daling et al, 1992;Frisch et al, 1997). A high proportion of anal tumours are associated with human papillomavirus (HPV) infection (Melbye and Frisch, 1998). HPV 16 DNA has been reported in up to 93% of invasive squamous anal tumours and 69% of anal carcinoma in situ (Tilston, 1997).Most epidemiological studies on HPV infection and anal cancer have been case -series and case -control studies using samples taken after the cancer has been diagnosed. Such studies may be subject to differential misclassification related to the presence of the disease, and provide no information on the temporal order of events. Prospective studies are generally considered crucial for causality inference. As primary prevention of HPV infection by vaccination is being evaluated, it is important to establish which cancers are likely to be amenable to prevention. Prospective seroepidemiological studies have linked HPV to vulvar, vaginal and to a subset of head and neck cancers (Bjørge et al, 1997a;Mork et al, 2001).HPV serology, using capsids of HPV 16, 18 and 33, has been validated as a type-restricted marker of past or present HPV infection (Dillner, 1999). A serologic association of HPV 16 with incident anal cancer has been shown (Heino et al, 1995). In a study of the role of HPV in non-cervical anogenital cancers, we found that HPV 16 seropositivity conferred an increased risk for noncervical anogenital cancers, but not for anal cancer (Bjørge et al, 1997a). This was in spite of the fact that the proportion of anal cancer cases (25%) being HPV 16 seropositive was higher than for other non-cervical anogenital cancers (24%). This...