This paper provides an analysis of the rural-urban divide in intergenerational educational mobility in Indonesia with two distinguishing features. First, the estimating equations are derived from theory incorporating rural-urban differences in returns to education and school quality, and possible complementarity between parent's education and financial investment. Second, the data are suitable for tackling the biases from sample truncation due to coresidency and omitted cognitive ability heterogeneity. The evidence rejects the workhorse linear intergenerational educational persistence equation in favor of a convex relation in rural and urban Indonesia. The rural-urban relative mobility curves cross, with the children of low educated fathers enjoying higher relative mobility in rural areas, while the pattern flips in favor of the urban children when the father has more than nine years of schooling. However, the rural children face lower absolute mobility across the whole distribution of father's schooling. Estimates from the investment equation suggest that, in urban areas, children~^!!^s peers are complementary to financial investment by parents, while the adult role models are substitutes. In contrast, separability holds in villages. Peers and role models are not responsible for the convexity in both rural and urban areas, suggesting more efficient investment by educated parents as a likely mechanism, as proposed by Becker et al. (2015, 2018). The theoretical relation between the intercepts of the mobility and investment equations helps in understand whether school quality is complementary to or a substitute for parental financial investment. This paper finds evidence of substitutability, implying that public investment to improve the quality of rural schools is desirable on both equity and efficiency grounds.