Estimates of environmental values are frequently required as inputs to cost-benefit analyses when evaluating alternative resource use options. One strategy to avoid the high cost of conducting empirical work when non-market values are involved is to use value estimates from an existing study and to transfer them to the context of interest (a practice known as ‘benefit transfer’). However, the transfer of values is subject to a host of potential errors and could lead to poor policy recommendations. This paper reports the results of an Australian Choice Modelling study that was designed to address the issue of benefit transfer. The Choice Modelling technique is amenable to benefit transfer because, unlike the Contingent Valuation Method, it produces values for resource use outcomes that can be ‘decomposed’ into component values associated with particular attributes of resource use change. These attribute values have the potential to be ‘reconstructed’ according to the scenario changes under investigation in the new policy context. In this study, tests are conducted to examine the validity of transferring estimates derived in a national context to different regional contexts and inferences are made about the impact that differing frames of reference and population characteristics have on value estimates.