Teachers’ own level of human capital development is commonly believed to be deterministic for the quality and effectiveness of their instruction and management in the classroom. Yet, there still exists an international debate on whether better educated teachers contribute to students’ cognitive development. Leveraging a random class-assignment subsample (N = 3436) from a nationally representative teacher-student linked dataset in China, this study reassesses the ongoing contention regarding the value of teacher education. By linking differences in teachers’ own educational attainment levels across different subjects of instruction to variation in seventh grade students’ Chinese, Math and English test scores using student fixed-effect models, this study quantifies the cognitive returns attributable to better educated teachers, in student learning terms. Findings show that teachers with at least a bachelor’s degree contribute substantially to student learning compared to those who are less qualified, by as much as 0.069 SDs or about two additional months of learning over a typical academic year. Additional sensitivity analyses suggest that this observed effect is robust to model specifications, and is consistent for students from different backgrounds.