Previous empirical studies of government have focused primarily on aggregate financial measures of size. This analysis examines a different measure: the share of public employment in total employment. In empirical tests on OECD countries for 1970 and 1980, the level of public employment is shown to be a clear reflection of voter demand for public goods, electoral support for socialist parties, and political response to the problems of unemployment. In contrast, variables that measure characteristics of the political process, such as degree of government decentralization, strength of interest groups, and voter participation in elections, are not statistically significant.