Spatial separation of species at sowing has been proposed as a means of managing interspecific competition in mixed swards. This study examined the effect of row configuration on persistence of lucerne (Medicago sativa L.) in pastures and pasture–cover crop mixtures at three sites in the Central West, and in pasture mixtures at three sites in the Riverina, New South Wales, Australia. Lucerne density, taproot diameter, groundcover, and spatial distribution relative to the original drill row were measured at all sites, and plant-available soil water and light interception during spring were assessed at some sites. Row configuration (alternate or mixed drill rows) did not affect lucerne persistence; however, where lucerne seed was concentrated in every third drill row, intraspecific competition led to increased lucerne mortality. This was estimated to occur at densities >28 plants/m drill row. A lucerne density of ~55 plants/m2 in every or alternate drill rows (at row spacings of 250 mm) would likely achieve maximum lucerne production in the semi-arid environments tested, subject to the chance event of favourable conditions in the period after sowing that would maintain that density (e.g. cumulative summer rainfall >100 mm and summer day degrees <2160°C in 2 years at Cowra). The presence of a cover crop in the establishment year reduced lucerne density by 39% compared with pasture only, regardless of row configuration. Changed row configuration did not reduce competition for light under a cover crop, but there was a small increase in available soil water of up to 4.9 mm in the 0–1.15 m depth, mainly during the first summer, where pasture was sown in alternate compared with mixed drill rows with a cover crop. Soil was drier in pasture-only treatments than those with a cover crop, attributable to increased lucerne density and lower levels of litter cover on the soil surface. Pasture species remained largely confined to the original drill row, especially in drier environments, highlighting the importance of narrower row spacings for pasture establishment. In addition, we determined a mathematical relationship between lucerne density and the non-destructive measure of basal frequency; this relationship could be applied in mature lucerne stands with densities ≤80 plants/m2.