Significant work time in the U.S. is lost each year due to worker absence, but evidence on the productivity losses from absenteeism remains scant due to difficulties with identification. In this paper, we use uniquely detailed data on the timing, duration, and cause of absences among teachers to address many of the potential biases from the endogeneity of worker absence. Our analysis indicates that worker absences have large negative impacts: the expected loss in daily productivity from employing a temporary substitute is on par with replacing a regular worker of average productivity with one at the 10th-20th percentile of productivity. We also find daily productivity losses decline with the length of an absence spell, consistent with managers engaging in costly search for more productive substitutes and temporary workers learning on the job. While illness is a major cause of absenteeism among teachers, we find no evidence that poor health also causes lower on-the-job productivity.