What is rational for administrators to do depends on the situations in which they work. Pressed for quick recommendations, they cannot begin long studies. Faced with organizational rivalries, competition, and turf struggles, they may justifiably be less than wholly candid about their own plans. What is sensible to do depends on the context one is in, in ordinary life no less than in public administration. The situation is a bit like that of the recent immigrant who kept stuffing a vending machine with quarters and getting pieces of pie from the tray below. To his friend, who found him doing this and yelled for him to stop immediately, he said, What's it to if I keep winning? The situation in public administration, planning, and policy studies more generally is a little less funny. The vending machines are erratic. They cost more. And the advice of friends has generally been unsatisfactory, falling into two broad clusters of commentaries. The long familiar position promises that you get what pay for: to solve problems must define the problem carefully, collect all relevant information, rank values, evaluate alternatives, and select the best strategy. The other old friend, one day a satisficer, the next day an incrementalist, suggests that you get what can see: to cross an intersection, do not look all the way across town. Neither of these commentaries helps much in practice. The rational-comprehensive formulation can be a recipe for failure; the incrementalist formulation could have us cross and recross intersections without knowing where we were going. This essay proposes a resolution to this quandary that lies at the very heart of administrative theory and practice. By assessing degrees of complexity in decision situations, decision makers may be better able to adopt a strategy to fit the situation at hand, thus avoiding impossible information-processing demands, on the one hand, and unsatisfactory, making do, settling-for-less strategies, on the other hand. In particular. this essay seeks to refine the practical