We used differences in state school policies as natural experiments to evaluate the joint influence of educational quantity and quality on late-life physical and mental health. Using US census microsample data, historical measures of state compulsory schooling and school quality (term length, student-teacher ratio, and attendance rates) were combined via regression modeling on a scale corresponding to years of education (Policy Predicted Years of Education [PPYEd]). PPYEd values were linked to individual-level records for 8,920 black and 14,605 white participants aged 45+ in the REasons for Geographic and Racial Differences in Stroke study (2003-2007). Linear and quantile regression models estimated the association between PPYEd and Physical Component Summary (PCS) and Mental Component Summary (MCS) from the Short Form Health Survey. Models examined interactions by race and included adjustment for sex, birth year, state-of-residence at age 6, and year of study enrollment. Higher PPYEd was associated with better median PCS (β=1.28; 95%CI: 0.40,1.49) and possibly better median MCS (β=0.46; 95%CI:-0.01,0.94). Effect estimates were higher in blacks than whites (PCS x race interaction-β=0.22; 95%CI:-0.62,1.05) (MCS x race interaction-β=0.18; 95%CI:-0.08,0.44). When incorporating both school quality and duration, this quasi-experimental analysis found mixed evidence for a causal effect of education on health decades later.