The coefficient of variation (CV) fails in the evaluation of grain drill performance due to the induced variation by random sampling. A new index designated as the coefficient of uniformity U-c for the evaluation of seed distribution uniformity was introduced. It was established based on the least absolute deviation criterion rather than on least squares. The aptness of the new index to evaluate the seed distribution uniformity was investigated by comparing its performance with the ubiquitous coefficient of variation. The study proves that the U, was less sensitive for data outliers than the CV. The results also indicated that a consecutive sampling scheme should be considered during the grain drill feeding system evaluation instead of randomly selected samples. (c) 2006 IAgrE. All rights reserved Published by Elsevier Ltd.