Background: The relation between parental involvement and student achievement has been of research interest for many decades. Although the idea of reciprocal processes between parent and child was proposed 40 years ago, very few efforts have been made to investigate reciprocal relations between parental involvement and student achievement. Aims: Using self-determination theory, this study investigated the longitudinal associations of the manner of parental involvement (i.e., autonomy-supportive or controlling) in children's academic problems with children's academic achievement. This study further addressed the recently intensely debated methodological issue of examining reciprocal relations by comparing a random-intercept cross-lagged panel model (RI-CLPM) with the traditional cross-lagged panel model (CLPM). Sample and Methods: A RI-CLPM and a traditional CLPM were applied to 5-year longitudinal data including 1465 secondary school students (M[subscript age] at T1 = 10.82 years, SD = 0.62). In both models, we controlled for students' gender, school type, socioeconomic status and cognitive ability. Results: The results show that the RI-CLPM fitted the data better than the CLPM. Trait-like stability was found for both forms of parental involvement and academic achievement. At the between-person level, controlling involvement related to lower achievement, whereas no correlation between autonomy-supportive involvement and achievement was found. At the within-person level, there were positive reciprocal relations between autonomy-supportive involvement and achievement, whereas controlling involvement was not associated with achievement. Conclusions: This study contributes substantially to the understanding of the relations between parental involvement in children's academic problems and children's academic achievement by simultaneously taking between-person differences and within-person processes into consideration.