The relative yield potentials of surface (SD) and buried (BD) methods of drip irrigation and furrow irrigation (F) of cotton were compared over four seasons. Drip-irrigated treatments were maintained at a deficit of 45 mm below the fully-irrigated soil water content, while F was irrigated when the deficit reached about 90 mm. Nitrogen fertilizer was applied weekly to drip-irrigated treatments during irrigation over the first half of the season and to F as one application before sowing. Emergence of cotton was slower in a SD system than for BD or F. This delay was associated with slower sorption of water from SD laterals in the furrow to seeds sown in the top of ridges. In two seasons there was heavier yield from SD and BD irrigation than from F; in one season the yields were the same; and in one season there was a small reduction in yield with drip irrigation. Average lint yields for cultivar Deltapine 61 over the four seasons were 1633, 1736 and 1676 kg ha-1 for SD, BD and F, respectively. The yield of cultivar Siokra was higher and more stable than Deltapine types under drip irrigation. Drip irrigation delayed maturity by two to nine days, an effect associated with reduced fibre micronaire when compared with F. The relative yield of each treatment was maintained at all positions along the 200 m length of field. Therefore both drip systems must have maintained a consistent output of water and fertilizer over the length of drip laterals. Fibre length and strength were generally not affected by method of irrigation. It was concluded that even though the drip irrigation system had a higher yield potential, the performance of the system did not justify the high economic outlay to grow cotton at this site, especially in cool or wet seasons. © 1990 Springer-Verlag.