Several prospective studies with invasive carcinoma as endpoint have supported Human Papillomavirus as a cause of cervical carcinoma. However, the largest study used seroepidemiology and did not analyse presence of Human Papillomavirus DNA in the subsequent tumour. Linkage of serum bank registries and cancer registries had identified 196 women with a registered cervical carcinoma after donation of a serum sample. For the present study, biopsies for 127 cases could be located, verified to contain invasive carcinoma and be amplified by PCR. Three control women who had remained alive and without cervical carcinoma during an equal length of follow-up had been matched to each of the case women and tested for HPV antibodies. Presence of Human Papillomavirus DNA in the tumours was analysed by general primer and type specific PCR. HPV16-seropositive women had a relative risk of 4.4 (95% CI: 2.2 -8.8) to develop cervical carcinoma carrying HPV16 DNA. By contrast, there was no excess risk for Human Papillomavirus 16-seropositive women to develop cervical carcinoma devoid of HPV16 DNA. Prediagnostic HPV16 seropositivity was strongly correlated with later HPV16 DNA positivity of the tumour (P50.001) and prediagnostic HPV18 seropositivity correlated with HPV18 DNA in the tumour (P50.03). The link between prediagnostic seropositivity and type of viral DNA in the cancer implies that the carcinogenic effect of infection with these viruses is dependent on persistent presence of type-specific viral DNA. The most important risk factor for development of cervical carcinoma is infection with the oncogenic types of human papillomavirus (HPV). A series of cross-sectional case-control studies, based both on HPV DNA testing of tumours (Walboomers et al, 1999), measurement of serum HPV antibodies (Dillner et al, 1995), or both (Nonnenmacher et al, 1995;de Sanjose et al, 1996) have found a consistent and strong association between HPV and cervical carcinoma.That the association between HPV and cervical carcinoma is causal is supported also by prospective studies, both with cervical intraepithelial neoplasia (Koutsky et al, 1992;Chua et al, 1996) and invasive cervical carcinoma as the end-point Shah et al, 1997;Vonka et al, 1999;Wallin et al, 1999). However, most prospective studies are based on limited numbers of cases. The largest study to date is a seroepidemiological study that contained 182 cases (Dillner et al, 1997). However, prediagnostic seropositivity and type of HPV DNA in subsequently occurring invasive carcinoma has not been previously performed. Combining seroepidemiological studies with HPV DNA analyses in prospective studies could be important for several reasons. For example, there are indications that prior or ongoing HPV infections may interact in a complex fashion in cervical carcinogenesis (Luostarinen et al, 1999;Silins et al, 1999). By comparing prediagnostic assessment of virus exposure with presence of virus DNA in the tumour, it could be possible to infer whether type-specific persistence of HPV is the predominant...