The National Summer Learning Project (NSLP) examined the implementation and effectiveness of voluntary summer learning programs developed by five school districts--Boston, Massachusetts; Dallas, Texas; Duval County, Florida; Pittsburgh, Pennsylvania; and Rochester, New York--and their local community partners. The study spanned three phases. The RAND research team (1) collected formative data for strengthening the five summer programs in 2011 and 2012; (2) examined student outcomes after one summer (2013) and after two summers of programming (2014 and 2015); and (3) examined student outcomes in spring 2017, at the end of three school years after the second summer of programming. This seventh report in a series summarizes the findings of this third phase in the context of earlier findings and offers implications for policy and practice. Overall long-term findings show that, by spring 2017, the academic benefits for high attenders decreased in magnitude and were not statistically significant--although when benchmarked against typical achievement gains at the same grade level, they remained large enough to be educationally meaningful.