In 2009, 95 farmers in the mixed farming zone of southern New South Wales (NSW), average annual rainfall 450–700 mm, were surveyed about their use of perennial pasture species. Survey responses indicated that, on average, 52% of land was under crop, 29% contained perennial pasture and 19% annual pastures. The proportion of land sown to perennial pastures and the species used differed with rainfall. Farmers identified concerns about the cost of establishment and poor survival of perennial pasture species as constraints to wider adoption. The survey also revealed that cover-cropping (sowing pasture species under the final grain crop in a cropping phase) was the dominant method of pasture establishment. Large-scale, on-farm participatory experiments were sown with the farm machinery, three at Ariah Park and one at Brocklesby in southern NSW in 2009 (annual rainfall 100 mm less than long-term average), and a further two experiments (one at each location) commenced in 2010 (annual rainfall >200 mm above average). These experiments compared the effect of cereal cover-crop sowing rate (standard rates used by the collaborating farmer and half of the standard rate) on the establishment of the perennials lucerne (Medicago sativa), phalaris (Phalaris aquatica), cocksfoot (Dactylis glomerata), and chicory (Cichorium intybus) sown in different mixes and rates with various annual legume species. The persistence and productivity of individual species were monitored for 2 years after sowing. Results indicated little or no effect of the presence of a cover-crop on the initial establishment of any of the perennials, but pasture species survival were severely affected by cover-crop sowing rates as low as half of the farmer practice (10 kg barley or 12 kg wheat ha–1) in 2009. Despite higher than average annual rainfall in 2010 and 2011, the residual effect of establishing pastures under a cover-crop in 2009 was poorer persistence and lower productivity by lucerne at the standard cover-cropping rate, and by phalaris, cocksfoot and chicory at all cover-crop rates, and an increased incidence of weeds. Similar responses to cover-cropping occurred between 2010 and 2012, even with the wetter establishment conditions in 2010, for phalaris, chicory and weeds, despite demonstration at Ariah Park that higher populations of individual perennial species could be achieved by doubling the sowing rate of pasture seed in 2010. Lucerne compensated for lower plant numbers by increasing herbage growth in response to rainfall, but phalaris could not and total pasture productivity over the first 2 years after establishment was greatly reduced by the use of cover-crops in both 2009 and 2010. Cover-cropping also reduced annual legume seedset, which could have implications for future pasture performance. Lucerne was the most consistently productive perennial pasture species evaluated regardless of establishment technique or climatic conditions.