Human papillomavirus (HPV) vaccination trials have shown high efficacy (VE) against high grade cervical intraepithelial neoplasia (CIN2/3). CIN2/3 is a surrogate marker of invasive cervical cancer (ICC). These lesions may spontaneously regress. Therefore, long-term follow-up is needed to determine the overall VE against ICC. Between September 2002 and March 2003, 1,749 16- to 17-year old women from Finland were enrolled in the randomised FUTURE trial of the quadrivalent HPV vaccine (Gardasil) with active follow-up for 4 years. Passive follow-up using the population-based Cancer Registry started 6 months after the active follow-up ended in 2007. A cluster randomised, population-based reference cohort of 15,744 unvaccinated, 18–19 year old women was established. We linked these cohorts to compare the incidence rates of CIN3 and ICC. Passive follow-up after 4 years resulted in 3,464, 3,444 and 62,876 person years of follow-up for the HPV vaccinated cohort, the placebo vaccinated cohort and the reference cohort, respectively. The number of endpoints with CIN3 or ICC identified were 0 and 0, 3 and 0, and 59 and 3 for the three cohorts, respectively. The corresponding incidence rates were 0 (95% confidence interval 0.0–106.5), 87.1 (95% CI 17.9–254.5) and 93.8 (95% CI 71.4–121), respectively. Our study shows that evaluation of the long-term efficacy post vaccination for the most stringent endpoints is feasible using cancer registries.