Unreliable rainfall during the crop growing season leads to a variable use of applied fertiliser nitrogen (N) by the crop, which may leave substantial fertiliser N residue in the soil. Residual effects of fertiliser N (0-150 kg/ha) applied to a succession of wheat crops over the period 1987-94 were studied in terms of increased crop returns ($A/ha) from fertiliser application and increased soil mineral N for the subsequent crop. In spite of the unreliability of wheat responses to applied N in this region, increases in financial returns over this sequence of crops suggest that a strategy of routine N application to wheat was highly profitable on this fertility-depleted soil. When increases in returns from 1 fertiliser application were summed over successive crops, financial returns generally increased with increasing rate of N applied up to the highest N rate (100 or 150 kg/ha). When N was applied to each successive crop, financial returns were similarly increased but applications >50 kg/ha were less profitable than rates <50 kg/ha. Increased financial returns for the 7 crops grown with conventional tillage increased by $A306/ha, $794/ha, $867/ha and $867/ha for fertiliser N applied at rates of 12.5, 25, 50 and 75 kg N/ha to each crop, respectively. Total N fertiliser costs for the 7 crops were $A63ha, $126ha, $253/ha and $380/ha. Increased financial returns of $608/ha and $962/ha were derived from applications of 25 and 75 kg N/ha to each of the 7 crops with zero tillage. When N uptake by wheat was reduced by water deficit, or where a longer fallow period created much higher nitrate levels, a single fertiliser N application of 75 or 150 kg/ha resulted in nitrate accumulated to 1.2 m depth in the following May. Where N was applied to each crop in the sequence, application of 75 kg/ha increased soil nitrate to 1.2 m in the following May, except in 1989 and 1990. The 3-crop sequence, 1988-90, placed high demands on soil N supplies, with high wheat yields (about 4.5 t/ha) and grain N contents (100-115 kg/ha) in 1988 but lower yields (>2t/ha) in 1989 and 1990. Consequently, low levels (46-63 kg/ha) of soil mineral N were apparently carried over for crops in 1989 and 1990 even where 75 kg N/ha was applied to the preceding crop. Subsequent recovery of financial losses, incurred in years of water deficit, made the routine application of 75 kg N/ha to fertility-depleted soils of this region profitable.