Despite the fact that the average American student is absent more than two weeks out of every school year, most research on the effect of instructional time has focused not on attendance but on the length of the school day or year. Student and school fixed effects models using Massachusetts data show a strong relationship between student absences and achievement but no impact of lost instructional time due to school closures. I confirm those findings in instrumental variables models exploiting the fact that moderate snowfall induces student absences while extreme snowfall induces school closures. Prior work ignoring this non-linearity may have mis-attributed the effect of absences to such snow days. Each absence induced by bad weather reduces math achievement by 0.05 standard deviations, suggesting that attendance can account for up to one-fourth of the achievement gap by income. That absences matter but closures do not is consistent with a model of instruction in which coordination of students is the central challenge, as in Lazear (2001). Teachers appear to deal well with coordinated disruptions of instructional time like snow days but deal poorly with disruptions like absences that affect different students at different times.