Traditionally, tax administrations have used taxpayer audit program resources principally to deter deliberate noncompliance, 'encourage' due care in the exercise of tax obligations, and recoup otherwise forgone Revenue. Usually, a corresponding perspective has been adopted by those policy analysts concerned with identifying and modelling the 'optimal' design characteristics of taxpayer audit programs. But, in the process, they have presumed that tax administrations make no efforts to learn from the results of taxpayer audits. Equally, most policy analysts have presumed that taxpayer audit programs can play no part in tax administrations' efforts to improve the underlying willingness and ability of taxpayers to comply with their tax obligations. One important adverse consequence of these presumptions has been that the taxpayer compliance behavior literature has paid very little attention to the major policy issue of how tax administrations should go about allocating their taxpayer audit program resources so as to best manage what the candidate has termed the risks to the Revenue. These risks are defined to be those which stem from the lack of information available to tax administrations about both deliberate and inadvertent noncompliance by taxpayers. In the empirical part of the taxpayer compliance behavior literature it is now well-appreciated that both the nature and the degree of taxpayer noncompliance can vary considerably across different categories (populations) of taxpayers. Thus the 'risks to the Revenue' can likewise vary considerably across taxpayer populations. In turn, different taxpayer populations may well require different program 'treatments'. Accordingly, it is now being recognised in the empirical literature that much more attention needs to be paid to the types of information tax administrations require about taxpayer noncompliance so as to better perform their central role. It is already widely agreed by policy analysts that a tax administration's central role is a broad one, and that it entails ensuring, as far as is practicable and socially desirable, all taxpayers pay the correct amounts of tax - preferably voluntarily. However, a tax administration's performance in this regard has to be assessed largely on the results obtained from taxpayer audit programs. There is therefore a need for a new conceptualisation (model) of what might constitute an 'optimal' taxpayer audit program, and which better captures both the various aspects of taxpayer noncompliance and the information requirements of tax administrations. The need for such a model has now become an urgent one. This is especially because a number of tax administrations, including the Australian Taxation Office, are no longer seeking to use taxpayer audit program resources principally in the traditional deterrence, 'encouragement' and Revenue-recovery mode. Instead, these resources are increasingly being used to help identify those taxpayer populations, as distinct from individual noncompliant taxpayers, which represent the greatest 'risks to the Revenue'. In turn, the results from taxpayer audits conducted in the 'high risk' populations are being used to help the tax administration determine the most appropriate strategies for improving future compliance in these populations. It will be argued in this thesis that the capture of these important strategic characteristics of modern taxpayer audit programs cannot be achieved by augmenting the existing deterrence-based models. A complementary model, more suitable for wider policy analysis, is therefore developed which can readily encompass these characteristics. This model is based on the construct of a budget-constrained tax administration seeking to manage the risks to the Revenue in (what is described in the organizational literature as) a Learning Organization environment, and where market segmentation techniques are drawn on when making inter-population resource allocation decisions. The policy value of this model is then demonstrated by applying it to both quantitative and qualitative data compiled by the candidate for the Business Audit Program of the Australian Taxation Office. In the process, a number of separate, substantive contributions are made to the program monitoring, program evaluation and taxpayer compliance behavior literatures. Collectively, these contributions provide support for the central argument of this thesis that both taxpayer compliance behavior researchers and policy analysts now need to pay much more attention to the information-gathering and strategic resource allocation challenges confronting tax administrations. The policy issues arising here, it will be argued, go to the heart of what constitutes successful and accountable tax administration, and (in turn) a 'high integrity' tax regime which is both efficient and equitable.