Elections for the Scottish Parliament and for all Scottish local authorities were held on the same day in 1999 and 2003, allowing a more detailed exploration of turnout variations than has been possible before. On both occasions, turnout in the two sets of elections was almost identical so that reported ward turnout at local level also indicates ward turnout in the parliament elections. Parliamentary turnout is analysed at both constituency and ward levels. The latter gives us greater confidence in the constituency results and also suggests some relationships which are not apparent at constituency level. The strongest determinant of turnout is consistently found to be the proportion of adults with a degree; religious affiliation, here analysed for the first time at the aggregate level, is also strongly related to turnout.