Pedagogical content knowledge (PCK) is an important target of science teacher knowledge assessment. Most studies that have assessed the PCK across a large sample of science teachers used a text-based approach to elicit and assess the more declarative and static form of teachers' PCK. Recently, small-scale qualitative studies have adopted a novel video-based approach to characterize the more situated and dynamic form of PCK underpinning a teacher's in-the-moment pedagogical reasoning. In this relatively large-scale mixed-methods study, these two approaches to assessing PCK were compared. Specifically, a counterbalanced design was adopted to compare the test scores of 147 Grade 7 biology teachers in a video-based PCK test with those of the same teachers in a text-based PCK test with the same content. A subgroup of teachers (N = 30) with different levels of PCK was interviewed to further understand their reactions to the test and how these might have affected their test performance. The results of paired-samples t tests showed that the teachers' PCK scores in the two types of test were significantly correlated, but their scores were significantly lower in the video-based test. In terms of teacher reactions, although the teachers considered both types of test to be fair, they regarded the video-based test as more authentic and more closely related to their job demands. The teachers reported putting more effort into the video-based test and feeling more immersed and engaged, but also experiencing greater cognitive demand and a higher level of anxiety. A multiple regression analysis revealed that the cognitive demand perceived by the teachers significantly predicted their PCK scores in the video-based test. The findings have implications for the valid design of video-based PCK tests.