An experiment, established on a cracking clay (Vertisol) at Emerald, central Queensland, studied the dry matter (DM) production, nitrogen (N) fixation, and water use of several potential ley-legume species over 4 seasons (1994–1997). Four ley legumes (siratro, Macroptilium atropurpureum cv. Siratro; lucerne, Medicago sativa cv. Trifecta; lablab, Lablab purpureus cv. Highworth; and desmanthus, Desmanthus virgatus cv. Marc) were compared with a pulse (mungbean, Vigna radiata cv. Satin), and grain sorghum (Sorghum bicolor) was included as a non-legume control. Overall, the annual legumes lablab (17.5 t/ha) and mungbean (13.4 t/ha) and the perennial siratro (16.2 t/ha) accumulated more DM than the perennials lucerne (9.6 t/ha) and desmanthus (7.1 t/ha). Lucerne produced little DM in its first year, but in later years had similar production to siratro and lablab. Desmanthus produced >4 t/ha of DM in the first year but barely survived during later seasons. Annual legumes grew faster and exhausted soil water more rapidly than the perennials. The perennials were able to extract more water from the soil than the annual legumes and sorghum, but were inefficient at converting small to moderate rainfall events (25–50 mm) into DM production. During the fallow following the growth of lablab and mungbean, nitrate-N in soil increased and was always greater at the time of re-sowing than for the perennial legumes and sorghum. Initially, the 2 annual legumes derived a high proportion (50% to >70%) of their above-ground N from fixation (%Ndfa) but this declined as the experiment progressed to low values (<13%) in the third and fourth years, reflecting increased supply of nitrate from the soil. In contrast, %Ndfa peaked at 72% for siratro and >90% for lucerne, and remained high (25–50%) throughout the experiment. N fixation rates were strongly negatively correlated with soil nitrate. Over the 4 years, siratro fixed 161 kg N/ha, lucerne 120, lablab 119, mungbean 78, and desmanthus 19 based on above-ground biomass. Mungbean had a net negative N balance (–80 kg N/ha) due to N exported in grain.