This dissertation uses qualitative and quantitative case study data to appraise the application of economic demand models in pluralistic low income settings. Primary data from Cape Town, South Africa, are used to interrogate three elements of the neoclassical paradigm: the budget constraint, utility function and underlying preference-sets, and hyper-rationality and the assumption of full information. The neoclassical budget constraint may be affected by changes in price or income. This dissertation explains how perceptions of quality, community context and fear of social sanction may affect the perceived price of service access. Community context and support networks also affect an individual's available financial resources although conventional demand models do not capture the income effect of payments made directly to providers by non-household members. According to ordinal utility theory, consumers must be able to rank various bundles of commodities according to the satisfaction they yield. This dissertation considers how perceived quality, fear of community sanction, illness type, awareness of different providers and the mismatch of supply with perceived need can affect the composition and ranking of those bundles. Finally, New Institutional Economics (NIB) utilises satisficing and less than full information instead of strict optimisation. This dissertation attempts to formalise the NIB contributions by describing common strategies for satisficing and considering how the household and community context of care seeking may effect service use. A number of empirical and methodological techniques are used to estimate demand functions and move theory into practice. Longitudinal health diaries are used to collect data on health service use, expenditure and self-rated health. The data are analysed using panel data methods. The findings illustrate that, without a better understanding of demand-side barriers to effective diagnosis and treatment, curtailing the spread and impact of communicable diseases in pluralistic, low income settings will be difficult if not impossible.