Understanding nitrous oxide (N2O) emissions from agricultural soils in semi-arid regions is required to better understand global terrestrial N2O losses. Nitrous oxide emissions were measured from a rain-fed, cropped soil in a semi-arid region of south-western Australia for one year on a sub-daily basis. The site included N-fertilized (100 kg N ha(-1) yr(-1)) and nonfertilized plots. Emissions were measured using soil chambers connected to a fully automated system that measured N2O using gas chromatography. Daily N2O emissions were low (-1.8 to 7.3 g N2O-N ha(-1) day(-1)) and culminated in an annual loss of 0.11 kg N2O-N ha(-1) from N-fertilized soil and 0.09 kg N2O-N ha(-1) from nonfertilized soil. Over half (55%) the annual N2O emission occurred from both N treatments when the soil was fallow, following a series of summer rainfall events. At this time of the year, conditions were conducive for soil microbial N2O production: elevated soil water content, available N, soil temperatures generally > 25 degrees C and no active plant growth. The proportion of N fertilizer emitted as N2O in 1 year, after correction for the 'background' emission (no N fertilizer applied), was 0.02%. The emission factor reported in this study was 60 times lower than the IPCC default value for the application of synthetic fertilizers to land (1.25%), suggesting that the default may not be suitable for cropped soils in semi-arid regions. Applying N fertilizer did not significantly increase the annual N2O emission, demonstrating that a proportion of N2O emitted from agricultural soils may not be directly derived from the application of N fertilizer. 'Background' emissions, resulting from other agricultural practices, need to be accounted for if we are to fully assess the impact of agriculture in semi-arid regions on global terrestrial N2O emissions.