In the last few years there has been considerable discussion of what labor unions can and do accomplish for their members, and also what effect these accomplishments have on the general functioning of the economy. Much of this discussion has been focused on the question of how much unions possess. This is undeniably an important matter. It is my belief, however, that it is secondary to the question of how unions use whatever they do have. A convenient point of departure for our discussion is the question of what it is, if anything, that unions are attempting to maximize or optimize in their demands for wages and fringes. As we conceive it, in their wage-setting activities, unions satisfice rather than optimize. That is, union leaders seek to satisfy the wage expectations of their members, so as to avoid dissatisfaction, but do not go looking for trouble with employers simply because it might be possible to obtain still more. Hence the determinants of member expectations, Ross's orbits of coercive comparison, play an active role in determining union wage demands. This is hardly a new story, but certain of its implications are not generally appreciated. As we show, first, the proximate objectives of union wage-employment policy differ substantively when there is unemployment (among the membership) and when there is not. Only in the presence of unemployment do unions tend to become job rationers a la Perlman; however, as the next part of this paper contends this is not a peculiarity of unions, but reflects a behavior pattern common to many institutions. One consequence of this argument is that the economic power exerted by unions over wages--though not necessarily the possessed varies with the unemployment of the membership. The implications of this for the definition and measurement of union power, and its effect upon wages, are then discussed. (Author's abstract courtesy EBSCO.)