The effect of 2 tillage practices (zero v. conventional), fertiliser application (nitrogen, phosphorus and zinc), and pulse–cereal rotation on changes in soil mineral nitrogen, plant-available water in the soil, grain yield and protein, and key soil fertility parameters (total nitrogen, organic carbon) in the Central Highlands of Queensland were examined between 1991 and 1998. Four pasture treatments (perennial legume, perennial grass, annual legume and legume–grass mixes) were included in January 1995, following previously unsuccessful attempts to grow lucerne and annual medics. The experiment was conducted as an opportunity cropping system on an open downs soil at Gindie that is representative of a large proportion (70%) of soils in the Central Highlands. Tillage practice did not affect the amount of mineral nitrate or the plant-available water content of the soil at planting, except in 1991 and 1998 when plant-available water content was higher under conventional tillage than zero tillage. However, zero tillage improved grain yield in 2 of 4 years (wheat in 1992; sorghum in 1996), increased uptake of nitrogen in every crop and produced greater grain protein levels in both wheat crops grown than conventional tillage. There were grain responses to nitrogen + phosphorus fertilisers (wheat in 1991 and sorghum in 1997). Grain protein was increased with applications of nitrogen regardless of whether phosphorus was added in 3 of the 4 crops planted. Sowing a pulse did not significantly increase grain yields in the following crop although it did increase soil mineral nitrogen at planting. Soil nitrate remained low in control (P0N0) plots (