In the Swedish educational system, teachers have the dual responsibility of assigning final grades and marking their own students' national tests. The Government has mandated the Swedish Schools Inspectorate to remark samples of the national tests to see if teacher marking can be trusted. Reports from this project have concluded that intermarker consistency is low and that teachers' markings are generous as compared to those of the external markers. These findings have been heavily publicized, leading to distrust in teachers' assessments. In the article, we analyze and discuss the remarking studies from methodological as well as substantive angles. We conclude that the design applied in the reanalysis does not allow inferences about bias in marking across schools or teachers. We also conclude that there are several alternative explanations for the observation that teacher marks are higher than the external marks: The external markers did not form a representative sample, they read copies with sometimes marginal legibility, and they used a different scale for marking than the teachers had used. The results are thus not as clearcut as suggested by the reports and media releases, which is because a school inspections logic rather than a research logic was applied in designing, conducting, and reporting the studies.