The long-term effect of lucerne use, in reducing drainage of water below the root zone and runoff (water excess), has not been examined in south-western Australia (Western Australia). The main aims of the paper were to determine how the long-term mean water excess was influenced by the proportion of lucerne in the rotation and the length of the lucerne phase in relation to soil type and location. A simulation model was used to compare scenarios, drawing on historical weather data from 1957 to 2001. Simulations were performed for 2 locations (high and low rainfall) and 2 soil types (high and low water-holding capacity).Lucerne significantly and rapidly (within 2–3 years) reduces the long-term mean water excess in rotations consisting of 2–4 years of lucerne followed by 1–4 years of wheat compared with continuous wheat. For every 10% increase in the percentage of lucerne years in the total rotation length, the mean water excess decreased by 17–20 mm (7–9%) at Kojonup (high-rainfall site) and 7–8 mm (8–9%) at Buntine (low-rainfall site) compared with the water excess associated with continuous wheat at each location. The proportion of lucerne in the rotation had a greater effect on the water excess than the effect of different soil types. Variation in the water excess due to variation in rainfall was greater than the reduction in water excess due to lucerne. This makes the decisions about when to grow lucerne to reduce water excess difficult if livestock enterprises are less profitable than cropping enterprises. The simulations show that lucerne mean yearly biomass ranges from 4.5 to 6.9 t/ha at Kojonup and from 1.6 to 4.7 t/ha at Buntine, depending on soil type and stage of lucerne in the land use sequence. It is worth considering that lucerne has the potential to reduce subsequent wheat yields with removal in autumn.