The extent of the benefits of improved river health remain uncertain. Quantifying these benefits is useful in prioritising policy investments. This study uses the Choice Modelling technique to estimate the value that households attach to attributes of improved river health. Data from a choice modelling survey supported by DSE Victoria are employed to elicit household preferences in a case study of the Goulburn River. Results from conditional and nested logit model specifications indicate that respondents hold positive values for higher levels of fish and bird populations and for increasing riverside vegetation. The standard Hausman test for Independence-from-Irrelevant-Alternatives (IIA) assumptions violations is found to give inconsistent results. The value estimates of the conditional and nested logit models are shown to be statistically similar indicating that testing for IIA violation may be more complicated than currently assumed thus raising questions about the efficacy of the more complex nested logit model.