The recent boom of fact-checking has raised scholarly interest in these activities. As opposed to elections, the domain of direct democracies has remained unexplored so far. This chapter studies the journalistic work in the selection and presentation of fact-checked claims about referenda and initiatives. More specifically, the empirical analysis examines the extent to which journalists include checkable claims and whether the rating scores of political actors vary according to partisan affiliation. It focuses on Switzerland, the paradigmatic case of direct democracy, by examining 107 claims analyzed by TA-Faktencheck, the country’s most ambitious fact-checking project. These claims were made by politicians in the framework of the political TV show, Arena, on 14 ballot propositions that were submitted to the people for voting purposes from February 2016 to May 2018. Two main findings emerge from a quantitative content analysis. First, slightly more than 40% of the included claims fail to pass the “checkability test” since they refer to opinions, predict the future, or are vague in nature. Second, this study documents considerable partisan variation in terms of rating scores, as the members of the Swiss People’s Party from the radical right are shown to perform significantly worse than those from the Social Democrats. This chapter suggests that journalistic fact-checking practice needs to be improved.