In 1995, the Flemish Government commissioned a two wave panel study on the citizens knowledge about the Flemish institutions, structures, and policy, and on the effect of an intensive information campaign held between the first and the second wave. 710 respondents co-operated in thefirst wave. In 1996 the study was repeated, following a simulated test-re-test design with 532 panel respondents and 455 new respondents. This study is based on the 987 respondents in the 1996 survey. After a discussion of the theoretical models for political knowledge, a model with eight latent variables and with a number of social-background variables is introduced and tested, using a structural modeling approach. The model has a strong predictive power for political knowledge (64% explained variance). The most important explanatory variables are education, and reading political news in newspapers. Other explanatory factors for the variation in political knowledge are generation, and membership of voluntary associations, gender, individualism, and reading local news. The 'consumption' of local news or sensational news about crimes or calamities has a negative effect on political knowledge. The lack of political knowledge clearly affects both trust in polities and the experience of a complex incomprehensible society. The less the knowledge the more distrust and feelings of uncertainty. These relationships clearly show the importance ofknowledge about politics for the quality of democracy.