It is both theoretically and practically important to investigate the types and features of parental involvement and their relationships with secondary school students’ subject literacy. Using the data of Hong Kong and Macao in PISA 2018, this study conducted a latent class analysis (LCA) to identify four different latent classes of parental involvement. We then used the robust three step method (R3STEP）and BCH respectively to test how covariates such as region, student gender, and grade contributed to the classification of parental involvement and how different classes of parental involvement were related to students’ mathematics literacy, reading literacy, science literacy, and global competence. The results showed that (a) region, student gender, grade, family ESCS, and parents’ expectation of their child’s educational level were significantly related to, while the gender of parents/guardians who filled the questionnaire was not significantly related to the classification of parental involvement; (b) students of different parental involvement classes were in general found to be significantly different from each other in terms of their subject literacy. Specifically, students of Class 1 (Low Involvement Class, C1, 39.1%) were found to have the highest average scores in mathematics literacy, reading literacy, and science literacy, but not in global competence. Students of Class 2 (Busy and High Involvement Class, C2, 33.3%) were found to have slightly higher average scores in reading literacy than those of class 3 (Comprehensive and High Involvement Class, C3, 23.1%), but have lower performance in mathematics literacy, science literacy, and global competence than those of Class 3. Students of Class 4 (Limited and Moderate Involvement Class, C4, 4.5%) were found to have the lowest average scores in mathematics literacy, reading literacy, and science literacy, but have better performance in global competence.