In recent decades, US voting has not only polarized but also nationalized: state-level voting choices increasingly reflect national allegiances. Such cross-level nationalization can undermine accountability. Yet understanding its sources – and whether it is distinctively American – requires comparative data. The authors assemble a novel data set linking 146,934 subnational and national election returns across 14 European and American democracies over several decades, supplemented with surveys. They find that nationalization has been steady and high for decades in most countries; the US is an outlier only in having nationalized recently. Their findings challenge monocausal explanations emphasizing media or public opinion. Instead, greater subnational authority and stronger subnational attachments predict lower nationalization. They also uncover ties between nationalization and affective polarization. More polarized voters are more likely to cast congruent ballots, though at the country level this link is driven by the US. This title is available as Open Access on Cambridge Core.