Spatial reasoning, the ability to mentally represent and transform objects and their relations, is considered so closely connected to mathematics that there is no longer a question of whether the two are related. Instead, there remains debate about how to answer the question of why. This paper explores the way the two fields of mathematics education and psychology define, assess and examine this important relationship. We find that both fields rely extensively on psychometric tests to define spatial reasoning skills and do not characterise spatial reasoning in context. We look at the points of connection and propose that the best way forward is to build on the complementary approaches currently undertaken. We suggest better alignment between the spatial and mathematics reasoning skills that have been identified (theoretically and for assessment development), and a focus on how spatial reasoning interventions influence how students think and learn. With the growing interest in spatial intervention for mathematics achievement, we analyse the different ways this has been undertaken and suggest directions for future research. For example, how might laboratory-based studies be scaled up within a range of classroom contexts, and how might classroom-based studies offer more control settings and systematic variation?