Background Vaccine effectiveness against COVID-19 needs to be assessed in diverse real-world population settings . Methods A cohort study of 805 741 residents in Scania county, Southern Sweden, aged 18-64 years, of whom 26 587 received at least one dose of the BNT162b2 vaccine . Incidence rates of COVID-19 were estimated in sex- and age-adjusted analysis and stratified in two-week periods with substantial community spread of the disease . Results The estimated vaccine effectiveness in preventing infection> 7 days after second dose was 86% (95% CI 72-94 %) but only 42% (95% CI 14-63 %) > 14 days after a single dose . No difference in vaccine effectiveness was observed between females and males . Having a prior positive test was associated with 91% (95% CI 85 to 94 %) effectiveness against new infection among the unvaccinated . Conclusion A satisfactory effectiveness of BNT162b2 after the second dose was suggested, but with possibly substantially lower effect before the second dose.