Accurate measurement of metacognitive knowledge in reading is important. Different instruments and scoring methods have been proposed but not systematically compared for their measurement comparability across cultures and validity. Given student data from 34 OECD countries in the Programme for International Student Assessment (PISA) in 2009, we compared two scoring methods for metacognitive knowledge in reading based on pair-wise comparisons of strategies and with conventional Likert-scale responses of selected items. Metacognitive knowledge scored with conventional Likert-scale responses demonstrated higher cross-cultural comparability than the pair-wise comparison method. Linked with reading competence, motivation and control strategy in reading, scores from the two scoring methods showed differential criterion validity, possibly related to the types of tasks (understanding and remembering versus summarising), item content (complexity and discrimination between preferred strategies in reading) and common method variance (e.g., individuals' stable response style in rating scales). Theoretical and methodological implications are discussed.