Using large, geographically representative surveys from the US and UK, this paper documents variations in the percentage of tasks workers can do from home. The authors highlight three dimensions of heterogeneity that have previously been neglected. First, the share of tasks that can be done from home varies considerably both across as well as within occupations and industries. The distribution of the share of tasks that can be done from home within occupations, industries, and occupation-industry pairs is systematic and remarkably consistent across countries and survey waves. Second, as the pandemic has progressed, the share of workers who can do all tasks from home has increased most in those occupations in which the pre-existing share was already high. Third, even within occupations and industries, the authors find that women can do fewer tasks from home. Using machine-learning methods, this paper extends the working-from-home measure to all disaggregated occupation-industry pairs. The measure that is presented in this paper is a critical input for models considering the possibility to work from home, including models used to assess the impact of the pandemic or design policies targeted at reopening the economy.