An interlaboratory study in which the insulin contents of five plasma samples were estimated in 36 italian laboratories was coordinated by the Istituto Superiore di Sanità (National Institute of Health) and the Consiglio Nazionale delle Ricerche (National Research Council). A rather large between-laboratory variability resulted, though the ranking of samples according to their insulin concentrations was practically the same. A significant dependence of estimates on the method used was established. The analysis of data, aimed at defining the possible reasons of the assay variability, is reported and discussed.