We examine the out‐of‐sample predictability of excess returns in the Australian government bond market. Our results confirm previous findings that a linear combination of forward rates provides a statistically significant prediction of bond excess returns on 1‐ to 5‐year maturity bonds. However, from an asset allocation perspective, our predictive model fails to obtain positive economic utility against the no‐predictability benchmark. Our results are robust to the sample period and different parameter assumptions.