This study takes a cross-country perspective to examine whether inaccurate teacher judgements of students’ math skills correlate with student social origin and whether such bias is associated with math achievement in primary school. We focus on England, Germany, and the US because these countries differ in the teachers’ growth mindsets, accountability, the use of standardised tests, and the extent of ability grouping. The data stem from three large-scale surveys, the Millennium Cohort Study for England, the National Educational Panel Study for Germany, and the Early Childhood Longitudinal Study, Kindergarten Class of 2010–11 for the US. At the beginning of primary education, teacher judgements were not entirely consistent with student scores in standardised tests. In England and Germany, teachers underrated students with low-educated parents and overrated those with high-educated parents. In the US, no such differences were found. In all three countries, overrated (or underrated) students performed better (worse) later on. In England and, to a lesser extent, in Germany, we found evidence that biased teacher judgements contribute—over the course of primary school—to widening inequalities in value-added achievement by parental education. Such effects were negligible in the US. Our findings suggest that a cross-country perspective is essential to better understand contextual factors’ role in systematic bias in teacher judgements and its relevance for educational achievement. This study can be seen as a starting point for future research to investigate the mechanisms of such contextual effects more thoroughly.