Abstract This article argues that the 'classical experimental' approach to researching adventure education has inherent weaknesses which have prevented it progressing towards understanding the process, rather than just describing the outcomes. Drawing on an approach to evaluation of crime reduction programs, an alternative research paradigm is offered for adventure education. Within this paradigm causality is understood as the interaction between the participant and the program. One implication is the use of a broader range of research methods, including participants' accounts of their experiences, which Barrett and Greenaway (1995, p.54) noted were 'almost entirely absent' from the research they examined in their review. A further implication of the 'new' research paradigm is that, in order to build on previous understandings of the process of adventure education, research should start from theory. The second part of the article offers this starting point, through a synthesis of previous work; including Mortlock (1984), Priest (1991), and Hopkins and Putnam (1993); and the author's own experience. This theoretical model of the process of adventure education provides a foundation for hypothesis generation. Introduction This article provides a model of the process of adventure education, drawing on previous research and theoretical contributions. The model provides a platform for further research and the generation of specific hypotheses. It helps to overcome the criticism that program evaluations have failed to build on previous work and it sets a framework in which to situate contemporary research. The article is mainly directed towards researchers in the hope that it will lead to more systematic research into adventure education. However, the model of the process of adventure education offers a new synthesis of previous work. The article proposes that the realist approach (Pawson & Tiliey, 1997) to evaluating adventure education will be more valuable than the classic experimental design, which has been used by most research so far. The differences between the two approaches are discussed. The realist approach offers the new research paradigm called for by Barrett & Greenaway (1995) in their review of previous research. An implication of the realist approach is that research should start from a theoretical model of adventure education; which it is the main concern of this article to present. The theoretical model helps distinguish adventure education from other activities, such as adventure activity courses, or management training. The model defines adventure education by its objective of personal growth; and the way this is achieved through personal adventure into new understandings of ones self. It offers a synthesis of previous understandings of the process by which this is achieved. Thus, the article advocates both a new approach to evaluating adventure education programs, and a theoretical model from which to start this. The model, and research proposals, are generic to adventure education, but the particular research hypotheses may be adapted to the Australian or United Kingdom contexts. Problems with previous research Problems with previous research are illustrated with reference to the simple evaluation model, figure one. This presents the elements of the process of adventure education in sequence. This model can be seen to illustrate the classic experimental design, associated with a positivistic approach (Pawson & Tilley, 1997). Typically research is looking for measurable outcomes of a program in an experimental group, and comparing them to a control group who have not experienced the same program. Causation between the program and intermediate effects, or the final outcome, is inferred from the repeated succession of similar effects after similar programs. A problem faced by research is proving a causal relationship between elements of boxes 1-5, figure 1, and isolating the effect of any one of them. …