The author presents new evidence on the effects of attendance on academic performance. He used a large panel data set for introductory microeconomics students to explicitly take into account the effect of unobservable factors correlated with attendance, such as ability, effort, and motivation. He found that neither proxy variables nor instrumental variables provide a solution to the omitted variable bias. Panel estimators indicate that attendance has a smaller but significant impact on performance. Lecture and classes have a similar effect on performance individually, although their impact cannot be identified separately. Overall, the results indicate that, after controlling for unobservable student characteristics, attendance has a statistically significant and quantitatively relevant effect on student learning. (Contains 5 tables and 22 notes.)