The problem addressed in this thesis is the development of strategic knowledge through the teaching and learning process. Specifically, it addresses strategic knowledge in three-dimensional solid modelling computer aided design (3DSM-CAD). The problem is addressed through two separate studies. In Study 1 the research question, What constitutes strategic 3DSM-CAD knowledge? is addressed. The outcomes of the first study are then used to inform the second study which addresses the research question: What instructional strategies will improve the use of strategic 3DSM-CAD knowledge? It is argued that CAD knowledge is inadequately theorised and a reconceptualisation of 3DSM-CAD knowledge is proposed. This proposed reconceptualisation is tested in Study 1 using a sample of 3DSM-CAD experts. The findings of this study are interpreted as supporting the theorised reconceptualisation of 3DSM-CAD knowledge. Study 2 investigates whether the use of expert strategies by novice 3DSM-CAD learners can be improved though the teaching and learning process. An instructional intervention is devised and empirically tested. The findings of this study provide support for the design of the intervention. It is argued in this thesis that the existing conceptualisation of CAD (computer aided design) knowledge is too narrow and is inadequately theorised, particularly the current interpretation of strategic knowledge. Presently strategic CAD knowledge is narrowly defined in terms of making choices between methods of model production. This thesis draws on existing research on expertise and in cognitive science to interpret strategic knowledge in terms of the metacognitive processes undertaken by experts when they engage in 3DSM-CAD processes. Strategic 3DSM-CAD knowledge is thus more broadly interpreted to include the metacognitive processes of planning, predicting, process selection, monitoring and evaluating. It is argued that this reconceptualisation is needed, as much of the existing research in CAD expertise is based on two-dimensional and three-dimensional CAD and not the parametric three-dimensional CAD currently used in industry and in education. The parametric nature of current 3DSM-CAD software provides a situation whereby decisions made at an early stage in the modelling process may later lead to model failure. The development of the use of expert 3DSM-CAD strategies is therefore considered important not only in order to improve the efficiency of the modelling process but also in ensuring the integrity of the model during subsequent design modification. Study 1 of the thesis addresses the delineation of 3DSM-CAD expertise. It uses existing research in expertise, the identification of experts and in knowledge elicitation as a basis for the study. In addition, a theoretical examination of the literature in cognition and mental imagery is used to conceptualise the 3DSM-CAD process. 3DSM-CAD experts are identified for case studies that investigate the cognitive processes undertaken by 3DSM experts as they engage in 3DSM-CAD processes. The selection of experts is undertaken using criteria identified in existing research in expertise: they have a number of years experience in the domain, they currently use 3DSM-CAD on a regular basis, others regard them as possessing domain expertise and, they have experience in teaching/training others. The findings of this study suggest that, while these existing criteria are useful when identifying experts in the use of complex computer applications, there is the possibility that these activities may be specific to the domain in which the software is used. For example, 3DSM-CAD capabilities may be specific in engineering or architecture but not to 3DSM-CAD in general. It is therefore argued that higher level experts may be identified by the additional criteria of their engagement in a range of problem solving activities across a range of contexts such as is found in 'Help Desk' personnel. Study 1 also builds on existing knowledge elicitation research through the development and testing of a knowledge elicitation technique, video-capture with think-aloud protocols, which enables the capture of very detailed data regarding the cognitive processes employed by individuals as they interact with complex computer applications. The findings of Study 1 support the proposed reconceptualisation of 3DSM-CAD knowledge and, in particular, the metacognitive nature of the strategic knowledge employed by 3DSM-CAD experts. Study 1 also found that 3DSM-CAD experts employ a range of mental imagery techniques, including those identified in this thesis as unique to 3DSM-CAD, in the process of solving 3DSM-CAD problems. Further, the findings establish that 3DSM-CAD experts engage in 'expert parsing' when dividing an object into parts prior to undertaking the modelling process. 'Expert parsing' is similar to the characteristic of experts in other fields who employ a deeper and more structured problem solving approach, looking beyond the surface features of an object. 3DSM-CAD experts identify deeper features and alternative algorithmic saliencies so that the choices of processes that allow both efficient object modelling and the possibility of making future changes may be made. 3DSM-CAD experts are found to bring a different approach to the 3DSM-CAD process based upon the 'encapsulation' (Boshuizen, Schmidt, Custers, & Van de Weil, 1995) of knowledge into an 'easy to change' script. Study 2 addresses the research question: What instructional strategies will improve the use of strategic 3DSM-CAD knowledge? Study 2 builds on the findings of Study 1 proposing that expertise in 3DSM-CAD involves both the development of metacognitive processes and 'expert parsing'. It is argued that, due to the perceptual nature of the processes involved in 3DSM-CAD, the metacognitive processes involved in expert 3DSM-CAD performance are facilitated by mental imagery. The development of the intervention employed in Study 2 is derived from the current research in metacognition, mental imagery and the design of learning strategies. A multi-treatment intervention is designed and tested empirically. This treatment is found to be effective in improving the use of expert 3DSM-CAD strategies. From these two studies it is concluded that the proposed reconceptualisation of 3DSM-CAD knowledge into three levels; declarative command knowledge, specific procedural command knowledge and strategic knowledge is appropriate. Further, the research supports the metacognitive nature of strategic knowledge and thus the need to address the development of expertise in 3DSM-CAD through a cognitive approach to instruction.