One of the most enduring puzzles in comparative politics is how Japan’s Liberal Democratic Party (LDP) has managed to stay in power, with limited interruptions, for nearly seven decades. We investigate the role of voters’ policy preferences and partisanship in perpetuating this dominance, with a conjoint experiment based on hypothetical party manifestos in the 2021 parliamentary elections. Our design features sets of party-choice exercises with and without party labels, and also asks respondents to identify which hypothetical manifesto is closest to the LDP’s. Our results show that: (1) the LDP’s policies are not the most popular, even among its supporters; (2) these policies become more popular when attached to the LDP label; and (3) the party label effect applies even for policies that are not perceived as being close to the LDP’s positions. These findings indicate that the LDP’s dominance should not be attributed to the popularity of its policies.