In recent years, one of the most significant pieces of environmental legislation in New Zealand has been the Resource Management Act (New Zealand Government 1991) that has empowered local government agencies to manage the use of natural resources in their regions. Three Government Departments have been responsible for developing policies directly relating to the use of natural resources in New Zealand. The Department of Conservation has been mainly concerned with the management of natural resources on public land. The Ministry for the Environment has particularly addressed environmental policy issues of national significance. The Ministry of Agriculture and Forestry has worked with New Zealand's agricultural, horticultural and forestry industries to encourage sustainable resource use and development for the benefit of all New Zealanders. In general, local and central government agencies carrying out policy analyses have drawn upon highly goal driven theories such as Rational Choice or Incremental Policy Theories or alternatively they have applied more loosely framed theories such as Systems Policy Thinking or Garbage Can Theory. Policy formulation and instrument selection may have been based upon instrumentalist, proceduralist, contingentist or constitutivist selection criteria, depending upon the assumed influence of peoples' behavioural and social contexts in addition to the technical characteristics of the tools themselves. However, there has been a limited range of policy theories to guide the integration of policy analysis, and formulation and operational planning into a management strategy for effective policy delivery. Such theories would have assisted policy agencies to identify the human and social behaviours most closely related to policy issues and to better match policies to differences in the political and social context of each of the issues that they were dealing with. In academic articles a number of behaviour models from social psychology have been used to explain and predict human behaviour. One of those, the Theory of Reasoned Action (TRA) has a long history of use in research and application. It has been adapted to suit the needs of policy makers in human health, marketing, and education. Applications of the TRA have been reported to have achieved coefficients of determination for behaviour of on average, 53% in one study and 71% in another. Some of the modified models based upon the Theory such as the Theory of Planned Behaviour, have in themselves been able to make additional contributions to peoples' understanding of how to explain and predict human behaviour in more complex situations. In this report, unless otherwise stated, references to the TRA are inclusive of all associated models, such as the Theory of Planned Behaviour. This thesis has examined the application of the Theory of Reasoned Action in the formulation of environmental policy. Five research questions were considered. 1. Could a human behaviour model based upon the Theory of Reasoned Action be developed sufficiently for environmental policy makers to explain landowner behaviour associated with managing indigenous vegetation? 2. How well could the social psychology model of human behaviour based upon the TRA have predicted public responses to a policy programme? 3. How well could the social psychology model of human behaviour based upon the TRA have distinguished between the policy-intervention needs of different stakeholder groups? 4. How much have peoples' values, attitudes and beliefs affected their behaviour? 5. What would be the immediate antecedents to peoples' behaviour and how have they led to behaviour change? This has been a quantitative study to develop and test models of human behaviour specific to the preservation of indigenous vegetation. Three data sets were compared from surveys of peoples' bush protection behaviour, the establishment of indigenous woodlots and the protection and planting of riparian areas with indigenous vegetation. The results from the analyses have shown that accounting for peoples' intentions could have been used to improve the estimates of peoples' use of policy-desired practices. The coefficients of determination in multivariate equations to predict peoples' natural resource behaviour based upon non-specific (external) variables, varied between 3 - 10%. By including intentions in the models, the level of explanation increased to 10 - 17%. The results may have been lower than expected from other examples in the literature due to poorly specified measures of behaviour relative to the measures used for intentions. When it came to estimating intentions (rather than the actual behaviours), the TRA variables in regression equations achieved coefficients of determination of 55 - 75% and these provided a measure of how well the underlying values, attitudes and beliefs could have given policy makers an understanding of peoples' behaviour. Comparing the beliefs of people with high and low intentions to perform the behaviours, clear differences have been identified that could have been the basis of policy strategies for behaviour change. After analysing and considering these examples, this thesis has argued that the TRA could be used in the future to provide policy agencies with an increased level of understanding of human behaviour and so enable them to formulate policy interventions for achieving predictable levels of behaviour change.