Abstract Background Although BNT162b2 vaccine-efficacy analyses have been published, the effectiveness of the vaccine in preventing coronavirus disease 2019 given confirmed exposure has not been previously demonstrated, even though it has policy implications, such as the need for self-quarantine when exposure has occurred. Methods In a retrospective cohort study, we used data collected between 20 December 2020 and 17 March 2021 from the second largest healthcare provider in Israel to analyze the probability of an additional household infection occurring within 10 days after an index infection. In model 1, vaccine effectiveness was described for Fully Vaccinated individuals (7 or more days from second dose) vs either Unvaccinated individuals or those Recently Vaccinated Once (0–7 days from the first dose, presumably still unprotected). Secondary analyses included correction for differing testing rates. In model 2, we conducted a separate analysis of households comprised of only adults with the same vaccination status. Results A total of 173 569 households were included, of which 6351 had an index infection (mean [standard deviation] age, 58.9 [13.5] years); 50% were women. Adjusted vaccine effectiveness of Fully Vaccinated compared with Unvaccinated participants was 80.3% (95% confidence interval [CI], 73.5–85.4) and 82.0% (95% CI, 75.6–86.8) compared with those Recently Vaccinated Once. Conclusions The BNT162b2 vaccine is effective in high-risk real-life exposure scenarios, but the protection afforded in these settings is lower than that previously described. Individuals with a confirmed significant exposure to severe acute respiratory syndrome are still at risk of being infected even if fully vaccinated.