This study investigated whether and, if so, how metacognitive knowledge can be assessed validly in students with special educational needs in a large-scale assessment like the German National Educational Panel Study. In total, 804 sixth-grade students including both regular school students attending the lowest track of secondary education (Hauptschule) and students with special educational needs in learning participated in the study. A scenario-based test of metacognitive knowledge focusing primarily on different aspects of strategy knowledge was implemented. In order to investigate optimal testing conditions, two conditions that varied in terms of administration mode were compared: "autonomous reading" as in regular test settings and a "read-aloud" condition. Reading speed and reasoning abilities were assessed as control variables. As expected, regular school students outperformed students with special educational needs in the metacognitive knowledge test. In addition, higher correlations between metacognitive knowledge and reading speed emerged in the autonomous reading condition compared to the read-aloud condition. Contrary to our expectations, a differential boost due to the testing accommodation of reading aloud was, however, only observed in regular students but not in students with special educational needs. The results are discussed with regard to educational and assessment-relevant approaches.