This article uses prefecture-level panel data from Japan, spanning the period 1989–2003, to examine the influence of social norms and fractionalization on voting behaviour. The key findings obtained from analysis via the fixed effects estimation, which controls for unobserved prefecture-specific fixed effects, are as follows: (1) the voter turnout is higher in close-knit communities, indicating that social norms enhance voting; (2) fractionalization, from both economic and generational standpoints, lowers the voter turnout and (3) a lack of social capital can lead to the distribution of votes being spread thinly among the competing parties.