Girls’ concurrence of higher educational achievement and higher depression levels seems incompatible with the negative correlation between educational achievement and depression. This seeming puzzle suggests hidden gender disparities and disconnections between various lines of research. I propose a framework of five correlation patterns to accommodate this seeming incompatibility, with each pattern representing a unique gender disparity structure. To test this framework, I employ structural equation models to analyze two waves of data from the China Education Panel Survey. The results support a “female-driven, male followed” pattern of correlations, i.e., compared with boys, girls have a stronger, negative correlation between achievement and depression. Moreover, the reciprocal relationship between achievement and depression holds only for girls, not boys, which partially explains girls’ stronger correlation between achievement and depression. For girls, achievement and depression contribute equally to their reciprocal relationship. Finally, the gender disparity in depression is more pronounced among low-achieving students. High-achieving girls do not exhibit higher levels of depression than high-achieving boys, while low-achieving girls exhibit much higher depression levels than low-achieving boys. Such results suggest low-achieving girls are particularly vulnerable to depression. These findings have broad theoretical implications, and the framework of five correlation patterns can also be adapted to investigate other forms of inequality.