A method of evaluating individual grass plants for seed shattering is described. The procedure involves harvesting a sample of culms from each plant, just prior to the start of shattering in the field, and maturing these detached culms indoors, sustained in tap water. A machine is described which is used to provide a uniform amount of shattering stress to each culm sample by means of a series of agitations of controlled severity. A mean coefficient of variation of 12.04 per cent was obtained using this technique on 32 clones of reed canary grass. The application of the technique to other species is discussed.