Policymakers view parental involvement (PI) as a crucial component of school reform efforts, but evidence of its effect on student achievement is equivocal. Using the Early Childhood Longitudinal Study-Kindergarten Cohort dataset, we examined the long-term impact on student- and school-level achievement of three types of school-based PI: PI to help an individual's own child (private-good PI), PI to improve the school (public-good PI), and PI through peer networking. Hierarchical linear modeling analyses indicated that all three types raised student-level achievement in mathematics but only private-good PI was associated with reading achievement. Public-good PI and parent networking boosted school-level achievement. Public- and private-good PI were more strongly associated with student-level mathematics achievement for high socioeconomic status (SES) students; aggregated private-good PI was more strongly related to school-level achievement in low-SES schools. These results provide empirical evidence about the effectiveness of school-based PI, but also suggest a need for schools to explore more effective ways to leverage the social capital of low-SES families.