Given the potentially harmful consequences of affective polarization on democracies, studies have predominantly focused on its causes. Recent research suggests that affective polarization tends to rise during elections, while it seems to decline afterwards, as the salience of political news reduces. Elections, as such, could become a double-edged sword: while they constitute the cornerstone of liberal democracies, they might actively fuel attitudes that harm them. Yet, we know little on whether the electorate as a whole polarizes during elections, or that this average increase is driven by a specific subset of voters. In this study, we speak to this research puzzle by asking the question: Which voters polarize in the run-up to the elections? Using insights from different strands of literature, we identify three potential subgroups of the electorate that are most likely to polarize: partisans, politically sophisticated voters, and voters with extreme ideologies. Empirically, we rely on a panel dataset that is fielded around the Belgian Federal elections in June 2024, which captures respondents’ levels of affective polarization (both vertically and horizontally) in three different waves (pre-campaign, campaign, post-election). This data allows to deal with the limitations of earlier research by capturing a baseline level of affective polarization before the kick-off of the election campaign. In this way, we contribute to the ongoing debate about the drivers of affective polarization.