Abstract. Nitrogen (N) fertiliser is a major source of atmospheric nitrous oxide (N 2 O), and over recent years there has been growing evidence for a non-linear, exponential relationship between N fertiliser application rate and N 2 O emissions. However, there is still a high level of uncertainty around the relationship of N fertiliser rate and N 2 O emissions for many cropping systems. We conducted year-round measurements of N 2 O emission and lint yield in four N-rate treatments (0, 90, 180 and 270 kg N ha -1 ) in a cotton-fallow rotation on a black vertosol in Australia. We observed a non-linear exponential response of N 2 O emissions to increasing N fertiliser rates with cumulative annual N 2 O emissions of 0.55, 0.67, 1.07 and 1.89 kg N ha -1 for the four respective N fertiliser rates, but no N response to yield occurred above 180 kg N ha -1 . The annual N 2 O emission factors induced by N fertiliser were 0.13, 0.29 and 0.50% for the 90, 180 and 270 kg N ha -1 treatments respectively, significantly lower than the IPCC Tier 1 default value of 1.0%. This nonlinear response suggests that an exponential N 2 O emissions model may be more appropriate for estimating emission of N 2 O from soils cultivated to cotton in Australia. It also demonstrates that improved agricultural N-management practices can be adopted in cotton to substantially reduce N 2 O emissions without affecting yield.