Light-duty vehicles emit ∼20% of net US greenhouse gases. Deployment of electric vehicles (EVs) can reduce these emissions. The magnitude of the reduction depends significantly on EV charging patterns and hourly power grid variations. Previous US EV studies either do not use hourly grid data, or use data from 2012 or earlier. Since 2012, US grids have undergone major emission-relevant changes, including growth of solar from ∼1 to ∼20% of generation in California, and >30% reduction of coal power countrywide. This study uses hourly grid data from 2018 and 2019 (alongside hourly charging, driving, and temperature data) to estimate EV use emissions in 60 cases spanning the US. The emission impact of charging pattern varies by region. In California and New York, respectively, overnight EV charging produces ∼70% more and ∼20% fewer emissions than daytime charging. We quantify error from two common approximations in EV emission analysis, ignoring hourly variation in grid power and ignoring temperature-driven variation in fuel economy. The combined error exceeds 10% in 30% of cases, and reaches 50% in California, home to half of US EVs. A novel EV emission approximation is introduced, validated (<1% error), and used to estimate EV emissions in future scenarios.