Two fields of cotton on two nearby farms in the Macquarie Valley of NSW which had different histories of weed management were studied to determine if weed seeds were being introduced via irrigation water. The experiment was set up as a randomised block design with five rows on each fann studied. At the beginning of each irrigation, traps constructed from wire gauze with holes of approximately 1.5 mm diameter were placed over the outflow end of each of the siphons feeding the selected rows. At the beginning of the season soil samples were taken from the furrows of the selected rows to determine the size of the weed seed bank in each field prior to irrigation.