Lacking national direction, in 2020 Brazilian states adopted a variety of nonpharmaceutical interventions to combat COVID-19, adjusting their stringency over time. We took advantage of this variation across states and over time to estimate the independent effect of seven interventions on the growth rate ratios of COVID-19 cases and deaths. Panel regressions using daily data from March to December 2020 for twelve states show that two interventions, suspension of public events and masking mandates, significantly reduced the spread of the disease. Partial and full suspension of public events were equally effective. Full masking mandates were more effective than partial ones. Together, suspension of public events and full masking mandates reduced the growth rate ratios of both cases and deaths almost to 1, the point at which cases and deaths are no longer increasing. Policy makers may find this analysis useful as they aim to choose the most effective nonpharmaceutical interventions to reduce COVID-19's social and economic burdens.