The connections between theory in human geography and theory in political science are sadly ill-developed. Of the three types of theory in political science-normative, descriptive and positive-it is the last two which would seem to have the greatest relevance for geography. Normative theory hardly features in geography since the subject has neither an explicit nor implicit normative base. Of the likely fields of political theory which would be fruitful in geography, those concerning individuals and their motives, and decision theory seem particularly promising. Of the positive theory, the theory of games at a micro-level and the theory of the firm at a macro-level seem to have great potential. Geographers should learn much from the political literature on decision-making in an environment of uncertainty. The new political geographer should be concerned with the decision forces affecting the distribution of space as a scarce public good and consequently with the values, the organization and the access to power of groups. THE decision to confront the theme of public policy at the I974 Conference must have concentrated a number of minds wonderfully, because there is still an astonishing lack of effective connection between theory in human geography and theory in political science. I would argue that the word 'astonishing' is justified, for the need for such connection is now so clear. Younger human geographers, working for their Ph.Ds, are tending to turn away from highly quantified macro-models because they find them lacking in depth of causal explanation and thereby (they feel) inadequate in predictive power. They are increasingly concentrating instead on micro-studies of spatial decision processes by one man or (more usually) a number of people and groups working in interaction: the decision to locate a factory in a development area, the decision to build an urban motorway on one line rather than another. These workers are having to forge a new body of theory on their own; yet, apart from the neo-Marxian framework provided by Harvey (I973) and Pahl (1970) (and, behind them, French writers like Castells (I972) and Lefebvre (I972)), virtually no interpretation of political science models is available in spatial terms. In these circumstances, my contribution can be no more than suggestive. It is clearly impossible to make a systematic review of the vast body of literature which is of potential relevance, and to try to interpret it in terms meaningful to the geographer; in any case, such a systematic review exists in the early chapters of Bauer and Gergen (I968). Therefore, I shall try first to make a very general taxonomy of the theoretical work; then, within this framework, to focus on some of the more important insights; lastly, to suggest in very tentative terms some geographical applications. THE MAIN THEORETICAL APPROACHES The simplest possible division, made in Bauer and Gergen's review as in a number of other sources, is between normative theory and descriptive theory. One may ask how people and groups should act in order to get the best results; or one may ask how people actually go about getting the results they do, whether admirable or not. But then, borrowing also from the language of economics, one may speak of a third type; positive theory. Like normative theory, this is deductive in that it starts from certain rules or premises. But like descriptive theory, it seeks to explain what will happen under certain conditions, not what should happen. Nevertheless, because it has an inherent logical structure, it provides an implicit guide to policy-makers in that it seeks to show them the consequences of alternative policy actions. This three-fold classification is I believe the most useful one. It is not always easy to decide where any particular theory belongs within it. Some reviews describe whole groups of theories-in particular,