There has long been interest in education on the issue of “transfer” – the extent to which students can apply what they have been taught in school to solve related but novel problems or tasks. More recently, attention in this literature has turned to understanding whether certain teaching approaches are more likely to lead to transfer, such as integrating new learning with existing knowledge and comparing multiple cases with the same under-lying structure. Using data on 280,000 students in the 2019 TIMMS study, we investigate whether maths teaching that uses this approach is associated with primary students being able to solve mathematics problems that are not included on their country's national curriculum. We find no evidence that it does, which underlines the challenges involved in teaching for near, let alone far, transfer of academic skills.