The project described in this thesis was designed to investigate HIV preventive behaviour in heterosexual adolescents from a social-cognitive perspective. The project consisted of three studies. The design of each study was guided by a theory of human decision making, the Theory of Reasoned Action (TRA), and its extension, the Theory of Planned Behaviour (TPB). In the first phase, an elicitation study was conducted in which characteristics of the population were identified. This study consisted of a series of semi-structured interviews in which participants' sexual behaviours, beliefs and normative influences were elicited. This information was required in order to enable studies two and three to target beliefs and behaviours relevant to the population being examined. In the second study, a large-scale survey of 417 university aged adolescents was conducted. This study was designed, primarily, to examine quantitatively the determinants of safe sex intentions and behaviours within the targeted population. Determinants of behaviour examined were those included in the TRA/TPB. Hierarchical multiple regression analyses were conducted to determine the amount of variance in safe sex intentions and behaviours accounted for by TRA/TPB variables. In addition, a number of methodological issues related to criticisms of the TRA/TPB in the context of prediction of sexual behaviours were investigated. These issues included the utility of examining preference for alternative safe sex strategies in accounting for non-use of condoms, the importance of considering behaviour change as a process rather than an outcome, the importance of heat of the moment decision-making in the prediction of condom use behaviour, and the importance of consistent specification of partner type in survey instruments examining condom use behaviour. The third phase of the project consisted of an intervention study designed to modify HIV-risk related beliefs and behaviours. Design and evaluation of the intervention were guided by the TRA/TPB. The intervention was conducted using 102 participants, randomly assigned to intervention or control conditions. Methodological issues examined in study 2 were further explored in this study. Overall, it was found that the Theories of Reasoned Action and Planned Behaviour provided a useful framework for understanding, predicting, and modifying HIV-related sexual behaviours. The theories were found to explain between 67 and 73% of variance in condom use intentions, and between 52 and 55% of variance in behaviour. The theories were also found to be applicable to other safe sex behaviours with 50% of variance accounted for in monogamy intentions, and 34% of variance explained in behaviour, and 40% of the variance in non-penetrative sexual intentions, but only 6% of variance in non-penetrative sexual behaviour explained. In addition, a behaviour modification intervention based on the theories was found to lead to change in both predictor variables and behaviours, relative to controls. These changes were maintained over a 6-month follow-up period. That is, a series of mixed factorial Analyses of Variance revealed that the percentage of encounters in which a condom was used was significantly greater for the four weeks prior to follow-up compared with the four weeks prior to the intervention. In addition, the number of 'slip-ups' into unsafe behaviours was found to be significantly less. Change in TRA/TPB predictor variables which accompanied these behavioural changes included increased perceptions of behavioural control, more positive attitudes towards sexuality and safe sex behaviours, and more positive normative perceptions. These results were seen to provide strong support for the theories of reasoned action and planned behaviour and for their relevance to the study of HIV-preventive behaviour. However, prediction of safe sex behaviours was found to be enhanced in most cases by consideration of the methodological issues explored in each study. Thus, it was found that consideration of safe sex strategies other than condom use and the type of partner with whom the behaviour was occurring were important in both the prediction and evaluation of efforts to modify condom use behaviour. In addition, prediction of condom use behaviour and evaluation of intervention effects were enhanced by the consideration of behaviour change as a stage-like process, rather than as an all or nothing outcome. Results relating to the importance of heat of the moment decision making were less clear, and further research is recommended to clarify the issue. Overall, these results have important implications for safe sex education, and should be considered in efforts to modify unsafe behaviour as well as efforts to evaluate the effects of this modification.