Policymakers are increasingly using data to understand how well their colleges and universities serve students by examining such outcomes as graduation rates. In this report, we harness longitudinal data systems in Connecticut and Virginia to demonstrate how student-level data can be used to measure graduation rates that compare each institution’s students with demographically similar students around the state. We find that student demographics and academic preparation explain a considerable amount of the differences in six-year graduation rates between four-year institutions. At two-year colleges, we find that the overall adjustments are smaller and that race or ethnicity and family income play a larger role than academic preparation. Finally, we show that these adjustments cannot be approximated by using only publicly available institution-level data.