In initial cross-section estimates using data from the 1991-94 British Household Panel Study, the authors find that union members had lower overall job satisfaction than non-union members, and public sector workers had higher satisfaction than private sector workers. Controlling for individual worker effects (sorting) using panel methods confirms the lower satisfaction of union members, but eliminates the higher satisfaction of public sector workers. These results suggest that unions do not simply attract the dissatisfied, as previously suggested. By contrast, the greater satisfaction expressed by public sector workers seems largely a consequence of sorting, with those who are more easily satisfied being drawn to the public sector. Additional analysis of particular aspects of satisfaction, including satisfaction with pay, the work itself, and relations with the boss, generally supports these conclusions.