In discussions on root research methods, so far the choice of sampling schemes seem to have been neglected topics. I this paper sampling schemes are discussed for various situations. it is shown that the schemes used traditionally for raw crops such as serials may give a bias of upto 30% on total root dry weight. More reliable sampling schemes are presented, as well as possibilities for correction of data from traditional sampling.available data of variation is root mass per auger sample are summerised to calculate the number of replicate samples required to detect differences (of given magnitude) between two means. The coefficient of variation of root weight per auger sample seem to be fairly constant at around 40% in grassland, with slightly higher value to deeper layer of soil.