If in the week prior to the survey a respondent has had more than one hour of paid employment, or has performed more than 15 hours of unpaid work, they would not be counted as unemployed. So, for example, a 45-year-old laid-off business executive I met last week, who had started mowing his neighbors' lawns and doing odd jobs for cash because he needed money to have his car repaired in order to go on any interviews for a full-time job, would be counted among the employed members of the labor force.
This is partly why tracking hours worked is preferred by some economists, but all data have their caveats.
2006-2011 archives for The Daily Dish, featuring Andrew Sullivan