A method of calculating confidence intervals of the “area of influence” of a weed plant, and of yield losses calculated from it, was developed. In a worked example using published data, the confidence intervals of the area of influence were found to be large. Yield losses calculated from this method were less precisely estimated than those from a more traditional additive density experiment. This limited evidence suggests that to give similar precision, the area of influence experiments may need to be at least double their present size. If this is indeed the case, published statements on the space, time, and effort advantages of the area of influence design will need to be treated with caution.