This thesis examines the potential of the Social Representation (SR) approach as an alternative to the traditional social cognition approach adopted in the study of cancer-related thought and behaviour. The traditional approach is critically examined. The main motivation for this thesis concerned the non-compliance of cancer-related health behaviour despite the huge investment into health promotion campaigns. Given the lack of any methodological orthodoxy for the study of SRs the core aim of the thesis was to explore the methodology appropriate for SR analysis in the area of cancer. In recognition of the complex nature of SRs a multi-method approach was adopted in which four different methods were evaluated. As the basis of this research Qualitative methods (Chapter 4) were used to explore the underlying rationale of the representations of cancer, and to inform the design of more quantitative instruments. Next an idiographic approach (Multiple Card Sort Procedure) (Chapter 5) was employed to examine the cognitive component of the SRs of cancer. The affective elements making up the SRs of cancer along with the cultural elements were then explored using a Metaphor Procedure (Chapter 6). The fourth method, a self-report Questionnaire (Chapter 7) was used to examine the attitudes, knowledge and emotional cognitions making up the SRs of cancer. An attempt was also made to identify shared representations using each of the methods employed using analytic procedures consistent with the data collected. Thus a qualitative analysis was carried out on the qualitative data, a scaling analysis on the idiographic data. Both the metaphor and the questionnaire data were analysed using a cluster analytic methodology (Fife-Schaw, 1993). Each method proved to have strengths and weaknesses. The questionnaire approach for example proved to be the most useful in examining the relationship between the representations of cancer and health behaviour, but the constraints of this method were shown by the idiographic technique. This work provided a foundation for the second part of the thesis. Using a questionnaire format (Chapter 8) the measurement tool for the SRs of cancer was refined on a sample of 510 respondents. Five factors were identified, IIInesslRestriction, Challenge, Symptom Focus, Cancer Control and Emotional Aspects. When operationalised into scales these factors proved to be highly reliable with values in excess of 0.75 making them sound measurement tools. Using a cluster - discriminant technique three shared representations of cancer were identified, an Ambivalent representation, a Positive Control representation and an Illness representation which proved to be differentially related to health behaviour. The positive orientation is an interesting one because it receives less documentation within the cancer literature. The research suggests that social cognition models are failing to fully account for the variance in health behaviour for a number of reasons. Perhaps most important is the insufficient attention paid to emotionally arousing qualities of cancer and the role played by socio-cultural factors. Lifestyle approaches, experience of cancer, and dominant conceptualisations in the media seem to be influencing the representations of cancer held. The role of individual differences within the formulation of SRs was then examined (see Chapter 8). The findings suggest that SRs are a product of both the social environment and individual psychological differences. Cancer-related thought and behaviour then may vary according to the social context as well as individual traits. The implication for health educators is that they cannot rely upon a superordinate representational 'norm' but must look at the mores and mileu of the target group in question.