When a range of weed densities is needed for competition experiments, one method of reducing existing populations is to remove all weeds in strips of specified length down the crop row. This technique also can be used to create different sizes of weed dusters. Two methods of sampling weed densities after such thinning were compared: unrestricted sampling (under which quadrats are placed randomly within the row) and restricted sampling (under which quadrats are randomly placed within a row under the restriction that they coincide with the beginning of an infested strip). The bias of estimators of weed density under each of these two approaches was derived and bias-corrected estimators of weed density from the two methods were compared on the basis of their variance. The variance under restricted sampling is less than or equal to the variance under unrestricted sampling so that, by a minimum variance criterion, restricted sampling using a bias correction is the better technique.