A general model of crop yield response to herbicide application is proposed. The model includes three components: the effect of herbicide dosage on weed density, the effect of surviving weed density on crop yield and the effect of herbicide directly on the crop. The model is used to estimate the response of wheat yield to application of diclofop-methyl to control ryegrass .(Lolium rigidum) in Australia. It is found that the competitiveness of ryegrass plants surviving treatment is reduced by the treatment and that the proportion of yield lost at a given ryegrass density is not independent of the absolute weed-free yield. The response function is used to calculate economic thresholds and optimal herbicide dosages for ryegrass control in wheat by diclofop-methyl.