Winter cleaning is the removal of grasses from pasture using selective herbicides applied during winter. We compared the effectiveness of an early (June) and late (July) winter cleaning with an early spring herbicide fallow (September), spring (October) herbicide and no disturbance of the pasture on development of the root disease take-all in the subsequent wheat crop. Experiments were done at 5 sites in the eastern Riverina of New South Wales in 1990 and 1991. The winter clean treatments reduced soil inoculum of Gaeumannomyces graminis var. tritici (Ggt) compared with the other treatments at all sites as measured by a bioassay, with reductions from the undisturbed treatments of 52'79% over 5 sites. The winter clean treatments also significantly reduced the amount of take-all that developed in the subsequent wheat crop by between 52 and 83%. The early and late winter clean treatments increased the number of heads/m2 at 3 and 1 sites, respectively. Dry matter at anthesis was increased by the winter clean treatments at 3 sites. Grain yield was increased by the winter cleaning treatments over the other treatments at the 4 sites harvested, with yield increases of the early winter clean over the undisturbed treatment from 13 to 56%. The autumn bioassay of Ggt was positively correlated with spring take-all and negatively correlated with grain yield of the subsequent wheat crop at each site. However, there was a significant site and site Ã— bioassay interaction so that the autumn bioassay could not be used to predict the amount of take-all that would develop.