In most parts of the world, mathematics is one of the core subjects at school. Better performance in mathematics well predicts later academic success. Hence, it is important to understand how best we can help children to learn mathematics. In this talk, I will demonstrate how my research in previous years tackled this question. My studies on children’s place-value concept (i.e., the understanding of the base-ten numeration system) showed that children who performed poorly in mathematics had not mastered place-value concept well in early years, and such concept could be assessed reliably by a novel task called Strategic Counting Task. Then I moved on to conduct a pioneer longitudinal study to examine Chinese-speaking children’s learning difficulties in mathematics. The study revealed that there were multiple underlying cognitive deficits, constituting different subtypes of mathematical learning difficulties. From there, my work further investigated the cognitive mechanisms behind mathematical learning in different domains, hoping to provide a fuller picture of the relations between specific cognitive skills (e.g., spatial skills) and mathematics. Together, these findings offer insights into how children learn mathematics and where their difficulties are. I will also discuss future directions of my research to further advance our knowledge of how best we can support children’s mathematical learning.