Estimates of the compensating surplus generated by changes in non-marketed environmental amenities can be estimated using stated preference valuation techniques. These are typically framed in terms of WTP tradeoffs, even if the situation of interest involves a property right vestment that calls for a WTA question. The differences created by the two questioning formats are explored in this paper using the results of two choice modelling applications. Both applications were framed on the potential for irrigation development and environmental losses in the Fitzroy River Basin, Central Queensland. The scenarios used in the applications differed only in that they used alternatively WTP and WTA questioning formats. The results indicate that robust models could not be constructed from either WTP- or WTA-based data sets when only two alternatives were used in the choice sets. In contrast, a strongly fitting model was derived from WTP-based data where three alternatives formed the choice sets.