Using data on American women and the health status of their children, this paper studies the effect of remote work on female earnings. Instrumental variables estimates, which exploit a temporary child health shock as exogenous variation in the propensity to work at home, yield an hourly wage penalty of 77.1 percent. Earnings losses together with positive selection, and alternative first stage regressions, suggest that task re-assignment or lack of social interaction are likely mechanisms. The estimates also have implications for the costs of social distancing during a pandemic and may be especially applicable when children must be temporarily quarantined.