In a banana irrigation experiment, 25, 44, 64 and 83mm of water, inclusive of rainfall, were given to a plant crop in one and two applications per week. At both frequencies the mean bunch weight was significantly greater at a rate of 44mm/week, but there was no significant effect on weight of fruit harvested per unit area. Neither bunch weight nor yield per unit area was significantly affected by higher rates of irrigation. There was some evidence to show that twice weekly irrigation gave a greater weight of fruit per unit area than once weekly irrigation. Yields of fruit were highest at a soil moisture tension of 30 to 40 centuries. Yields of fruit and application fates of water were converted to increment ratios and fitted to third order polynomial curves.