Two experiments were conducted on plots sown in autumn 1992, at Tamworth in northern New South Wales. The first compared the establishment of 3 perennial grasses when sown as monocultures or with competitors in either broadcast-sown swards or alternate row-sown plots. Sowing rate and species of competitor were also examined as factors affecting perennial grass establishment. Perennial grasses were Austrodanthonia richardsonii (synonym Danthonia richardsonii) cv. Taranna, A. bipartita (synonym D. linkii) cv. Bunderra, and Phalaris aquatica cv. Sirosa. Competitors were Trifolium subterraneum var. brachycalycinum cv. Clare, T. repens cv. Haifa, and Lolium rigidum cv. Wimmera. In spring 1992, competitors were removed from 144 of the 288 plots to prevent them from seeding. A second experiment compared the longer-term (1993–96) dry matter yield and persistence of these perennial grasses under continuous grazing in plots were the competitor was present in year 1 (1992) or in all years (1992–96). In spring 1992, mean dry matter yield of perennial grass was higher (P<0.001) in row-sown plots then those sown by broadcasting. Mean dry matter yield of perennial grass was lowest at low sowing rate, but not significantly different at medium and high sowing rates (about 350 kg DM/ha). Compared with the monocultures, the presence of a competitor reduced mean perennial grass dry matter yields by 48, 69 and 85%, respectively for white clover, subterranean clover and annual ryegrass. Perennial grass plant numbers were highest (P<0.001) in the medium and high sowing rates of the monocultures and in white clover competitor plots and lowest (P<0.001) in all broadcast-sown plots, where annual ryegrass was the competitor. By spring 1996, white clover and annual ryegrass had declined to a low level in the pasture in all plots and the only major competitor was subterranean clover (1200 kg DM/ha, 40% plant frequency). Mean dry matter yields were highest (P<0.001) for Sirosa in 1993, but with dry conditions in 1994 and continuous grazing they were highest for Bunderra in all other years. The implications of these data for devising sowing strategies to maximise the establishment of perennial grasses and their long-term persistence in this environment are discussed.