The real-world effectiveness of the coronavirus disease 2019 (COVID-19) vaccines in Japan remains unclear. This case–control study evaluated the vaccine effectiveness (VE) of two doses of mRNA vaccine, BNT162b2 or mRNA-1273, against the delta (B.1.617.2) variant in the Japanese general population in the period June–September 2021. Individuals in close contact with COVID-19 patients were tested using polymerase chain reaction (PCR). A self-administered questionnaire evaluated vaccination status, demographic data, underlying medical conditions, lifestyle, personal protective health behaviors, and living environment. Two vaccine doses were reported by 11.6% of cases (n = 389) and 35.2% of controls (n = 179). Compared with controls, cases were younger and had a lower proportion who always performed handwashing for ≥20 s, a higher proportion of alcohol consumers, and a lower proportion of individuals living in single-family homes or with commuting family members. After adjusting for these confounding factors and day of PCR testing by multivariate logistic regression analysis, the VE in the period June–July (delta variant proportion 45%) was 92% and 79% in the period August–September (delta variant proportion 89%). The adjusted VE for homestay, hotel-based isolation and quarantine, and hospitalization was 78%, 77%, and 97%, respectively. Despite declining slightly, VE against hospitalization remained robust for ~3 months after the second dose. Vaccination policymaking will require longer-term monitoring of VE against new variants.