Background: Parental involvement in homework is a home-based type of involvement in children's education. Research and theory suggest that it is beneficial for learning and achievement under certain conditions and for particular groups of individuals. Aims: The study examined whether different types of parents' involvement in homework (autonomy support, control, interference, cognitive engagement) (1) are predicted by their mastery and performance goals for their child and their beliefs of the child's academic efficacy, and (2) predict student achievement goal orientations, efficacy beliefs, and achievement. Grade-level differences were also investigated. Sample: The sample consisted of 282 elementary school (5th grade) and junior high school students (8th grade) and one of their parents. Methods: Surveys were used for data collection. Structural equation modelling was applied for data analysis. Results: (1) Autonomy support during homework was predicted by parent mastery goal, parents' control and interference by their performance goal and perceptions of child efficacy, and cognitive engagement as supplementary to homework by parent perceptions of child efficacy. (2) Parental autonomy support, control, and interference were differentially associated with student mastery and performance goal orientations, whereas parent cognitive engagement was associated with student efficacy beliefs. (3) The structural model was the same for elementary and junior high school students but the latent means for a number of variables were different. Conclusion: Different types of parental involvement in homework were associated with different outcomes with parent autonomy support to be the most beneficial one.