Comparing the analyses based on the data of 1,235 Chinese children referred for government services and subsequently diagnosed as children with dyslexia in Hong Kong and those of 690 Chinese children in the sample for the normative study of the Hong Kong Test of Specific Learning Difficulties in Reading and Writing, we explored the reading–writing connection through a series of regression and correlation analyses. Specifically, orthographic knowledge, naming speed, and phonological memory were found to be salient predictors predicting both reading and writing, and when they were controlled and partialed out, the correlations between reading and writing were sizably attenuated, suggesting that these cognitive abilities or skills could act as third-variables contributing to the relationship between reading and writing. Issues related to causal inference, directionality of causality, and implications of the findings for intervention efforts and future research are discussed.