Purpose: The recent move towards fair-value accounting (FVA) indicates the belief of accounting standard setters that fair value information is decision useful, and hence meets the objective of general purpose financial report (GPFR). This paper aims to examine the usefulness of fair-value earnings through investigating its ability to explain and predict future cash flows using a sample of Australian agriculture firms. The paper also examines the difference in the abilities between fair-value measurement inputs. Originality & Key literature: Whereas the literature shows theoretical supports on the explanatory and predictive abilities of FVA, empirical evidence is scant. This paper contributes to the literature by providing empirical evidence on the usefulness of FVA. Methodology: Multiple linear regression and out-of-sample tests are undertaken to examine the explanatory and predictive power of fair-value earnings. Findings: Our empirical analyses yield two findings. First, fair-value earnings have ability to explain future cash flows only when it is measured by Level 3 (mark-to-model) inputs. Further, fair-value earnings show no predictive power for future cash flows regardless the measurement inputs adopted. Research implications: The results reveal that FVA does not satisfy the objective of GPFR in terms of information usefulness. Thus, the underlying belief to adopt FVA is not supported.