The paper investigates the effect of teacher characteristics (experience, level of education attainment and credentials, gender) on the achievement of eighth grade students using matched student-teacher data from the 1999, 2003 and 2007 TIMSS surveys. The analysis covers European countries, thus contributing to the existing exclusively US-based literature. Non-random student-teacher matching is present in these countries, as well, similar to the US, and this can lead to biased estimates of teacher effects. We deal with this problem using a rich set of student level controls on the one hand, and applying a student fixed effect specification on the other. The results overall suggest that the lack of experience and a proper teacher education degree might drive down student achievement. However, there is a substantial heterogeneity among countries; teacher characteristics seem to matter only in less than half of the countries analysed. Even within this group the results are not robust in some cases, as the fixed-effect estimates are often non-significant. Nevertheless, where these are present, the teacher experience and credential effects are similar in magnitude to those in the US studies reporting strong effects. For Hungary the impact of teacher characteristics is ambiguous; significant only in some specifications.