The British Standard for testing aggregates, BS 812:1989, requires at least ten scoopfuls of sand to be taken from any stockpile to obtain a representative bulk sample. This does not necessarily provide a representative sample, however, as the number of increments should be related to the variability of the batch being sampled. This paper describes trials in which increments were taken from stockpiles of sand and subjected to two tests: determination of fines content by washing and sieving and assessment by methylene blue titration of harmfulness of fines. It was found that the minimum number of increments required to provide a representative sample could be as low as two or as high as seventeen. The approach could be extended to other combinations of aggregate grades, test methods and specification limits.