The second rounds of elections to the Czech Senate consistently suffer from the lowest voter turnout among all types of elections in the Czech Republic. Moreover, voter turnout and the share of invalid votes decrease substantially between the first and the second rounds of the Senate elections. This article enquires into the causes of the decrease. It builds on theories that emphasise the salience of elections and the decisiveness of voting and using data from all Senate electoral races between 1996 and 2012 tests several theoretical predictions drawn from these theories. Two types of regression models are employed to test these hypotheses. Surprisingly, the findings show that neither the closeness of the electoral race nor the presence of an incumbent or a communist candidate has an influence on voter turnout. The decrease in turnout between rounds and between electoral cycles is best explained by the fact that first-round elections coincide with other types of elections, and by the lack of ideologically polarised electoral races in the second rounds and the fragmentation of the first round contest. The substantial number of invalid votes in the fi rst rounds can be attributed to voting error rather than deliberate electoral behaviour. The article closes with several suggestions regarding the electoral system used in the Czech Senate elections and with a discussion of the implications of the findings for current research.