The purpose of this thesis is to apply non-market valuation techniques to estimate the effect of coastal development and erosion protection on beach recreation values. The study area is the Coromandel Peninsula in New Zealand. The peninsula is a popular holiday destination and the coastal landscape generates strong feelings of attachment in many New Zealanders. However, twin pressures of rising coastal land values and shoreline fluctuations have led to conflict between people who want to protect or develop built environments and those who want to conserve or restore natural landscapes. The recreation amenity values which are under threat are difficult to include in a cost-benefit analysis of coastal policy because they have no explicit market price. I review the issues and limitations associated with non-market valuation methods and apply advanced visualisation and spatially-referenced data collection techniques to estimate the non-market values. This thesis comprises four papers which are either published or in the process of being published. A particular focus is whether 3D visualisations or "virtual environments" can improve the reliability and validity of stated preference results. The first paper describes the rationale and method for developing the virtual environments for use in a choice experiment about erosion protection and headland development. I find the virtual environments reduced bias, improved choice consistency and made respondents more likely to complete the survey. In the second paper I report the results of a choice experiment about development options for a specific undeveloped beach. I find that the virtual environment presentation format more strongly influences stated preferences amongst respondents with no direct experience of the site. The additional information provided by 3D visualisations may therefore be useful when people have to make decisions about the unfamiliar. In the third paper I report the results of tests of stability of stated preferences over time. A novel feature of this study is the use of two re-tests over time rather than just one. I find that stability at an individual level is positively associated with choice certainty. The virtual environments have a positive effect on respondent confidence but respondent education level has a larger effect. The large variation in stated value over time is consistent with the constructed preference viewpoint. However, the results are not necessarily incompatible with the alternative discovered preference hypothesis because there may have been too little consequential feedback to facilitate any preference learning. In the fourth paper I report the results of a destination choice analysis for beach recreation on the Coromandel Peninsula, using data collected in the same survey as the choice experiments. The focus of the fourth paper is the issue of spatially correlated errors caused by the spatial distribution of sites. Visitors are influenced by opportunities available at other sites and many visit multiple beaches in one trip, which violates the conventional assumption that sites are substitutes. I analyse the cumulative attraction of each pairwise combination of sites and review modelling approaches that allow for flexible patterns of substitutions but are also computationally efficient. I find that an Agglomerating and Competing Destination Choice (ACDC) model with differentiated accessibility parameters for each attribute offers the best fit. It has complex response properties for specific site changes yet retains a computationally feasible closed form. I use the model to examine the implications of two site-specific changes. The results highlight the importance of preserving coastal camping grounds and natural sand dunes because these attributes increase the diversity and attractiveness of the wider area as well as the individual site. The findings of these four papers and the data collected for this thesis make a significant contribution to knowledge about the recreation value of Coromandel beaches. This research is of particular relevance to the local policy issues of coastal erosion and development.