Specific memory B cells and antibodies are a reliable read-out of vaccine efficacy. We analysed these biomarkers after one and two doses of BNT162b2 vaccine. The second dose significantly increases the level of highly specific memory B cells and antibodies. Two months after the second dose, specific antibody levels decline, but highly specific memory B cells continue to increase, thus predicting a sustained protection from COVID-19. We show that although mucosal IgA is not induced by the vaccination, memory B cells migrate in response to inflammation and secrete IgA at mucosal sites. We show that the first vaccine dose may lead to an insufficient number of highly specific memory B cells and low concentration of serum antibodies, thus leaving vaccinees without the immune robustness needed to ensure viral elimination and herd immunity. We also clarify that the reduction of serum antibodies does not diminish the force and duration of the immune protection induced by vaccination. The vaccine does not induce sterilizing immunity. Infection after vaccination may be caused by the lack of local preventive immunity because of the absence of mucosal IgA.