Despite the evidence for high returns to education at an individual level, large increases in education across the developing world have brought disappointing returns in aggregate. This paper shows that the same pattern holds in India by building aggregates from micro-data so that the comparability and quality issues that plague cross-country analyses are not a problem. In India both men and women with more education live in households with greater consumption per capita. Yet in aggregate, comparing across age cohorts and states, better educated male cohorts consume only about 4% more than less well educated ones. Better educated female cohorts do not live in households with higher consumption. This result is robust to: (1) using econometric models that account for survey measurement error, (2) different measures of household consumption and composition, (3) allowing returns to differ by state, and (4) age mismeasurement. Comparing state returns to a measure of school quality, it does not seem that poor quality is responsible for the low returns.