The conventional wisdom among journalists and politicians is that higher turnout would benefit Democrats, although extant scholarly research suggests otherwise. We adopt a new approach to assessing the partisan impact of higher turnout. We use state-level exit polls and Census data to estimate the partisan preferences of nonvoters in Senate elections and then simulate the outcome of these elections under universal turnout. While nonvoters are generally more Democratic than voters, the dearth of close races means that very few election outcomes would have changed had everyone voted. Other scenarios—full turnout among registered voters, equal turnout rates for whites and African-Americans, and equal turnout rates across income groups—generate similar results: although Democrats fare better in each scenario, few outcomes would have changed. However, the gap between voters and nonvoters' partisan preference varies considerably across states and across years, suggesting that this “partisan differential” warrants further examination.