We examine the effect of air pollution on school absences using unique administrative data for elementary and middle school children in the 39 largest school districts in Texas. These data are merged with information from monitors maintained by the Environmental Protection Agency. To address potentially confounding factors, we adopt a difference-in-difference-in-differences strategy that controls for persistent characteristics of schools, years, and attendance periods in order to focus on variations in pollution within school-year-attendance period cells. We find that high levels of carbon monoxide (CO) significantly increase absences, even when they are below the federal air quality standard. Our work suggests that the substantial improvements in CO levels in the air over the past two decades have yielded economically significant benefits.