The media, newspapers especially, has in several studies been associated with high levels of political knowledge. Those whose preferred source of political information is newspapers show higher levels of political knowledge compared with those who prefer other information sources. In this article we claim that studying the effects of media on political knowledge is not a meaningful exercise unless a differentiation between structural knowledge and political information is made. The results of the analysis carried out in this article support this argument. Newspaper readership seems to have a moderate impact on political information but not on structural political knowledge. In general, mass media loses most of its relative importance as a predictor when political interest, age, education and gender are included. The results suggest that if we do not differentiate between different types of knowledge when measuring political knowledge, we lose a useful discriminant.