The authors analyze matched administrative survey data on economics students enrolled in two econometrics courses offered in consecutive terms at a major public university in Spain to assess the impact of lecture attendance and study time on academic performance. Using proxy variables in a cross-sectional regression setting, they find a positive and significant effect of attendance and study time, with a substantially higher return on each additional hour of attendance. However, when panel data first-difference estimators are used to eliminate time-invariant individual-specific unobservables possibly correlated with regressors of interest, the attendance effect disappears, while study time substantially increases its economic impact. These results suggest that study time may be much more important than attendance as a causal determinant of academic performance.