Partisan dealignment is recurrently presented in the literature as a main driver of the “personalization of politics”. Yet, on the one hand, the claim that leader effects on voting behaviour are increasing across time is short on comparative evidence. On the other hand, there is limited empirical evidence that such increase is due to dealignment. This article addresses these claims, exploring the longitudinal relationship between dealignment and the determinants of vote choice through a novel dataset pooling 90 national election surveys from 14 Western European parliamentary democracies in the period 1961-2016. The results suggest that both critics and proponents of the personalization thesis got it partially right. Leader effects did not increase over time, but their relative importance did: leader images came to matter more as party attachments came to matter less. Partisan dealignment is the key contextual dynamic in downplaying the electoral impact of partisan attachments vis-à-vis leaders evaluations.