The effects of rate of sowing, herbicide, and superphosphate on seedling establishment of two oversown improved grass species, their sward performance, and the resultant soil phosphate levels were studied in two experiments at Glen Innes, New South Wales in 1973. The experimental area had received 1250 kg ha-1 superphosphate prior to sowing and was carrying native grasses and white clover (Trifolium repens). In the first experiment, treatments applied were combinations of: four sowing rates (5.5, 11, 22, 44 kg ha-1) ; three herbicide rates (0 + 0, 8.3 + 0.55, 49.5 + 3.3 kg a.i. ha-1 2,2-DPA 4- amitrole) and two superphosphate rates (0, 250 kg ha-1). In the second experiment, three rates each of herbicide (0 + 0, 8.3 + 0.55, 16.6 + 1.1 kg a.i. ha-1 2,2-DPA + amitrole) and superphosphate (0, 125, 250 kg ha-1) were applied in a factorial design. Both experiments were sown with Phalaris aquatica and Festuca arundinacea. For both species, seedling numbers increased as herbicide and sowing rates increased but declined with superphosphate application. These main effects were more pronounced during the earlier samplings. Seedling numbers of both species, particularly phalaris declined with time. As herbicide rates increased, total dry matter production was progressively reduced, but production of fescue increased from 19 to 232 and phalaris from 0.5 to 60 kg dry matter ha-1. The addition of superphosphate eliminated the adverse effect of herbicide on the legume component. Superphosphate increased fescue dry matter production but decreased that of phalaris. Prior application of herbicide at the heaviest rate and superphosphate applied at 125 kg ha-1, both increased soil phosphate by approximately the same amount (7.5 and 8.2 p.p.m. respectively).