The aim of this study is to investigate the assessment practices of civics teachers in upper secondary school. The main purpose is to analyze, using Bloom’s revised taxonomy, which kind of knowledge civics teachers test in their written test examinations. The analyses show that they mainly test factual and conceptual knowledge. Procedural knowledge is tested in other ways, mostly by means of essays. Metacognitive knowledge is not found in this study. In contrast to earlier findings the teachers in this study test a variation of cognitive processes. Between 50 and 90 percent of test questions relate to the category remember, but there are also questions testing the categories to understand, analyze and evaluate, only a few test apply and create. There is a good alignment in teachers’ knowledge of the grading criterions and the curriculum. There are however discrepancies between this knowledge and their testing practices, which causes some problems concerning the validity of their tests. As the tests mainly are used for summative purposes, teachers varying practices lead to problems with grade equality. Frame factors may explain differences in practices. Mainly administrative factors such as working hours and schedule are significant, since teachers need time to prepare and to mark the tests and pupils need time to write them. Pupils also wish to have written tests and teachers adapt to this. Still, the significance of these factors is decided by teachers’ freedom of action. Most teachers know how to and want to make valid tests, but they need the time, both to prepare and to mark them, and the possibilities to extend lessons when more writing time is needed.