The alms of the current study were, firstly, to examme factors predicting nurses' perception of risk of HIV contagion at work; and secondly to investigate nurses' compliance with universal precautions, and the reasons for their compliance and noncompliance. The research was carried out in three phases. Phase I of the research used the Yates and Stone (1992) model of risk as a theoretical framework with which to examine perceived risk and also investigated reasons for non-compliance. The results suggested that perceived risk was related to fear of contagion and knowledge of non-transmission modes. Additionally four factors were produced from the investigation of reasons for non-compliance. Phase 2 utilised a qualitative analysis of semi-structured interviews designed to yield salient beliefs and issues relating to: risk perception and fear of contagion; knowledge of HIV; precautionary compliance; and the use of social comparisons when evaluating risk. The third phase of the research project was designed to combine the results of the first two phases of research with established theory in order to investigate the two main aims of the study. The Yates and Stone model of risk was again used. The Theory of Planned Behaviour (Ajzen, 1986) was used as a framework with which to examine compliance with universal precautions, and the False Consensus Effect (Ross, House and Green, 1977) was also used to examine behaviours and beliefs relating to both risk perception and precaution use. As with phase 1 this phase was questionnaire based. The relationships between descriptor and outcome variables were examined through the use of univariate, bivariate and multivariate statistical techniques. Overall it was found that fear of contagion, knowledge of HIV, previous exposure to HIV, and perceived behavioural control were all significant predictors of perceived risk of HIV contagion, results which support the Yates and Stone conceptualisation of risk. Four factors relating to precaution non-use were found and the Theory of Planned Behaviour was found to predict and explain precaution non-use accurately for 71 % of participants. Relationships amongst the variables were investigated in detail and the results were discussed in tenns of both theoretical and practical outcomes.