Previous research has that proportional-representation systems produce higher levels of government spending than do majoritarian systems. In some studies, scholars use district magnitude to reach this conclusion, showing a positive relationship between it and government expenditure. Yet, Carey and Shugart [1995: Electoral Studies 14, 417] argue that the effect of district magnitude on government expenditure is, in reality, conditioned on the level of incentives for personal voting in the electoral system. We find support for this hypothesis with an empirical analysis of government expenditure in 77 democracies between 1970 and 2000. 2006 Published by Elsevier Ltd.