In this article, electoral personalization is measured on the basis of the percentage of voters casting a preference vote instead of a list vote. A multilevel regression analysis of the 2006, 2010, 2014 and 2018 Lower Chamber elections in Colombia, with 971 open lists at the constituency level as units of analysis, shows that the electoral personalization has remained remarkably stable, in spite of the party system changes during the investigated period. Even so, leveraging the unique feature of the Colombian system that parties can opt for either a closed or an open ballot format, we find that the personalization of the open lists decreases with the number of closed lists, but only in the small constituencies. At the list level, personalization increases with the number of candidates on the list. When this effect is controlled for, we find evidence of a proximity effect in the sense that a larger number of voters in the district coincides with a lower level of personalization.