Levels of agreement between nine pathologists on the Rye classification of Hodgkin's disease and on diagnostic subcomponents used in applying the classification, were analysed by kappa statistics. Pathologists experienced comparatively little difficulty in agreeing on the presence of nodules and lacunar cells and hence best agreement was achieved on the nodular sclerosis category. Poorer agreement levels on the lymphocytic predominance, mixed cellularity and lymphocytic depletion categories were explained mainly by problems in the assessment of numbers of lymphocytes and abnormal reticulum cells other than Reed-Sternberg cells. Identification of the Reed-Sternberg cell, although of paramount importance to a diagnosis of Hodgkin's disease, appeared to have no great practical relevance to use of the Rye classification in this series of cases.