Nitrogen (N) fertiliser is almost universally used in high-yielding irrigated cotton, but it is not used efficiently in many instances. Predicting the economic optimal amount of N fertiliser is difficult and often little N fertiliser is required where situations have provided access to N through excessive N fertiliser being applied to previous cotton crops, conditions promoting significant N mineralisation, or if legume rotation crops were grown. The economic optimum N fertiliser rate (Nopt – where the marginal cost of N fertiliser (at $1.50 kg^–1 N) equalled the return on cotton lint (at $2.20 kg^–1) was determined in eight experiments conducted over 8 years; Nopt ranged from 0 to 248 kg N ha^–1, lint yields ranged from 1.3 to 3.4 t ha^–1, crop N uptake ranged from 96 to 321 kg N ha^–1 and apparent N fertiliser recovery (calculated by dividing the difference in crop N uptake between N-fertilised and unfertilised plots by the N fertiliser applied) ranged from 20% to 98% of N applied. A positive response to N fertiliser application in lint yield was evident in 7 of the 8 years. Both lint yield and crop N uptake were positively correlated with pre-sowing soil nitrate concentration. Cotton that yielded 1.4 t lint ha^–1 derived 78% of crop N from the soil, whereas at 3.4 t lint ha^–1, 69% of crop N was derived from soil; this indicated the importance of N supplied from the soil and the relatively lesser reliance on the N fertiliser applied, even for very high-yielding cotton. A multiple regression model, using the parameters of pre-sowing soil nitrate, crop N uptake and lint yield, more accurately represented the data generated in this study in estimating the economic optimum N fertiliser rate (r² = 0.80).