In a stated preference valuation survey, the expected benefits of environmental policies are generally presented to respondents without reference to the fact that the predicted outcomes are rarely known with certainty. This omission may reduce the credibility of the valuation scenario and contribute to hypothetical bias. In the study outlined in this paper, a choice experiment was conducted to elicit values for environmental improvements in the Great Barrier Reef (Australia), with outcome certainty included as a separate attribute. Different specifications of the utility functions, which imply different behavioural assumptions about the way choices have been made, generate variations in value estimates. Results showed that respondents incorporate outcome certainty into their decisions, but that an expected utility formulation, represented by the interaction between environmental protection and likelihood of occurrence, underestimated environmental values. Some environmental protection values appear to be independent of outcome certainty, which may be consistent with existence values and other non-use categories. A partial expected utility model is cautiously recommended.