The present study validated the Preschool Reading Attitude Scale (PRAS) using a longitudinal sample of young Chinese children in Hong Kong (N = 197; age: mean [SD] = 52.38 [3.32] months at the first wave of assessment). Children were asked to rate their own attitudes toward reading in a Likert format with pictorial aids, and received individual assessments of literacy skills. Parents rated children's interest in reading activities. The results showed that a two-factor model featuring institutional reading attitude and global reading attitude fitted the data the best. The two subscales demonstrated acceptable internal consistency reliabilities and satisfactory convergent correlations with parent-report child reading interest. In terms of the relation between attitudes and performance, child-reported institutional reading attitude was positively associated with receptive vocabulary and word reading concurrently and longitudinally. Moreover, this study provided evidence for the stability of the PRAS and weak measurement invariance across the two-year period. The results suggest that the PRAS is a reliable, valid, and developmentally suitable instrument.