Soil emissions of nitrous oxide (N2O) are generally low in Australian semi-arid cropping systems, but can be reduced further by incorporating legumes into cereal-based rotations. We used automated and manual chambers to compare N2O emissions throughout a two-year legume-wheat field experiment. Two pulse crops [lupin (Lupinus angustifolius L.) and field pea (Pisum sativum L.)] and two forage legumes [vetch (Vicia sativa L.) and clover (a mixture of four Trifolium spp.)] were grown in the first year, followed by a wheat (Triticum aestivum L.) crop in the second year without additional nitrogen (N) fertilizer under either no-till or tillage systems. All legume crops were either (a) chemically terminated at anthesis (brown-manured, BM), or (b) cut for hay (forage legumes) or harvested for grain (pulses) (product-removed, PR). The fate of legume N in the BM treatment was traced using 15N labelled urea applied to soil in micro-plots. Results showed that N2O emissions during the legume and wheat phases were mostly unaffected by legume species, the exception being lower emissions from the BM clover pasture under the no-till treatment. Under the PR treatment, tillage tended to increase N2O emissions compared with the no-till treatment during both legume (121 vs. 104 g N2O-N ha−1 year−1) and wheat phases (91 vs. 79 g N2O-N ha−1 year−1). In contrast, a mixed result was found under the BM treatment with a significant tillage × crop management interaction during the legume phase, but no effect of tillage during the wheat phase. In general, the BM treatment promoted more N2O emissions due to more N input into soil compared to the PR treatment, with the impact extending into the subsequent wheat crop. The N2O emissions from the BM treatment during the legume (195 g N2O-N ha−1 year−1) and wheat phase (181 g N2O-N ha−1 year−1) were greater than those from the PR treatment for the corresponding phases (113 vs. 85 g N2O-N ha−1 year−1). The 15N study showed that the majority of legume-derived N was retained in the soil at the end of the legume phase and likely to be readily available for the subsequent cereal crop. The mitigation of N2O emissions derived from legumes in a semi-arid cereal-based cropping system can be optimized in a no-till system where product removal is practiced (cut for hay or harvested for grain) with legume species choice having little or no additional impact.