Forty-one assays were analysed at the 3rd International Workshop on the standardisation of islet cell antibodies. Analysis of precision demonstrated assays consistently detecting blind duplicates within one doubling dilution and capable of discriminating one doubling dilution differences in islet cell antibody concentration. Some assays, however, reported duplicates discrepantly by more than seven doubling dilutions, and consequently could not distinguish even large quantities of islet cell antibodies. Precision was best in assays from laboratories which had participated in all three Standardisation Workshops and was not dependent upon methodology. The use of the Juvenile, Diabetes Foundation reference islet cell antibody standard and standard curves reduced the scatter of results, and was best amongst assays with better precision. Twenty-seven assays reported all ten blood donor sera as negative. However, 14 assays did not, and specificity (negativity in health) was less than 50% in three assays. Low specificity was strongly associated with poor precision. The detection limit of assays ranged from less than 5 to 50 JDF units and was partially dependent upon methodology. Assays incorporating extended incubation had the lowest detection limits without a decrease in the specificity of the ten blood donor sera. Precise quantification is fundamental for the standardisation and comparability of islet cell antibodies. Precise quantitative assays have been identified and reference standards and common units established.