Mass vaccination is effective in reducing SARS-CoV-2 infections among vaccinated individuals. However, it remains unclear how effectively COVID-19 vaccines prevent people from spreading the virus to their close contacts. Using nationwide administrative datasets on SARS-CoV-2 infections, vaccination records, demographics, and unique household IDs, we conducted an observational cohort study to estimate the direct and indirect effectiveness of mRNA-based COVID-19 vaccines in reducing infections among vaccinated healthcare workers and their unvaccinated household members. Our estimates for adults imply indirect effectiveness of 39.1% (95% CI: −7.1% to 65.3%) two weeks and 39.0% (95% CI: 18.9% to 54.0%) eight weeks after the second dose. We find that the indirect effect of mRNA-based COVID-19 vaccines within households is smaller for unvaccinated children than for adults and statistically insignificant. Here, we show that mRNA-based COVID-19 vaccines are associated with a reduction in SARS-CoV-2 infections not only among vaccinated individuals but also among unvaccinated adult household members in a real-world setting.