This study investigates the knowledge that legal practitioners use to solve authentic practical legal problems in naturalistic settings. The study examines the declarative and procedural knowledge that practitioners use in that context and whether experienced practitioners use knowledge organised in encapsulated and script form (Boshuizen & Schmidt, 1992; Schmidt, Norman, & Boshuizen, 1990) to enable ‘expert’ performance. The purpose of the study is to provide an empirically-based understanding of the knowledge used in solving real-life practical legal problems, for the information of the providers of practical legal training in Australia and other common law countries. The providers of that training use assumptions about that knowledge and how it is acquired, which do not always rest on coherent theoretical or empirically-derived foundations. The study uses the lawyering literature to identify the knowledge such literature considers is required to solve practical legal problems in lawyer and client interview settings. The study also examines the assumptions about the nature of that knowledge, and how it is acquired, which are apparent in the approaches of the providers of practical legal training. The limitations of those assumptions are identified from a cognitive perspective. The study examines cognitive conceptions of the knowledge used in problem solving in particular fields and how that knowledge becomes proceduralised and organised into structures called chunks and schemas. A particular examination is made of cognitive theories developed in the field of medical problem solving, which use the concepts of ‘encapsulations’ and ‘illness scripts’ to explain ‘expert’ performance in diagnosing disease in clinical settings. This analysis is used to synthesise the prediction that experienced legal practitioners may develop and use structures similar to encapsulations and illness scripts in problem solving. This prediction is based on the similarities between the way medical practitioners and legal practitioners are educated and trained, and are taught to solve problems using a hypotheticodeductive method (or a domain variant in the case of law), and on the similarities between clinical settings and lawyer and client interview settings. The study also examines theories that explain human problem solving by reference to a metaphorical ‘problem space’, and synthesises the prediction that practical legal problem solving can be explained by a problem space theory that was developed to accommodate complex, ill-defined problems. That theory uses the concepts of a problem zone to reflect the ill-defined nature of the problem as presented to the problem solver, a search and construction zone to reflect the phenomenon that the problem solver will have to construct operators to use to solve the problem, and a satisficing zone to reflect the phenomenon that there will be no single unambiguous solution to the problem (Middleton, 1998). The study uses the lawyering literature to identify the characteristics of practical legal problems in a lawyer and client interview setting. The cognitive literature is used to identify the cognitive conceptions that correspond to those characteristics. It is argued that these problems are complex, ill-defined problems that have to be found by the problem solver using weak problem solving strategies such as problem decomposition, attribute identification and means-ends analysis (Simon, 1973; Dillon, 1982; Newell, 1980). Based on these predictions two research questions are developed as follows: How do legal practitioners find and construct practical legal problems? Are there differences in the knowledge that experienced legal practitioners use and that which novice practitioners use? Do those differences reflect differences in the individual practitioner’s underlying knowledge and how that knowledge is organised? These questions are investigated in four case studies. Two of these studies involve experienced legal practitioners and two involve novices. These studies reveal that all the subjects used similar general problem solving strategies to find and construct problems. The subjects all constructed a series of problems rather than one large problem. The subjects did not all find and construct the same problems and some subjects’ constructions of problems changed as new information came to light. Most subjects did not finish the construction of problems at the interview. The processes that the subjects use to construct problems can be explained by Middleton’s (1998) problem space model, although this study suggests that model needs to be modified to accommodate the on-going emergent character of practical legal problems as they occur in lawyer and client interview settings. The investigation revealed qualitative differences between the problem attributes and moves that the experienced subjects identified and those that the novices identified. In summary, the experienced subjects identified attributes and moves that were more detailed, more directly related to the ‘facts’ and more concrete than those that the novices identified. Both the experienced subjects and the novices appeared to rely on recognition (Newell & Simon, 1972) to identify problem attributes and moves rather than on any apparent step-by-step legal analysis and reasoning process. This study suggests that the superior performance of the experienced subjects may be explained by their use of knowledge in encapsulated and script form, as predicted. The study discusses the implications of its findings for practical legal training courses as a need to provide students with general problem solving knowledge, provide them with the knowledge that they will need to recognise problems in specific areas of practice, to help them start to develop knowledge in encapsulated and script form, and to develop an understanding of the limits of institutional training in developing professional expertise.