Recent large-scale epidemiological studies demonstrate that blood concentrations of immunoreactive insulin predict the development of NIDDM and IDDM and are associated with the risk of several degenerative diseases, such as coronary and peripheral vessel atherosclerosis, hypertension, and dyslipidemia. The reliability of these measurements is dependent on a biological assay that has not been well standardized between laboratories. Recognizing this, the American Diabetes Association organized a task force to assess comparability of blood insulin measurements between laboratories and to suggest techniques to improve comparability. The task force found that identical serum and plasma samples measured in different laboratories produced widely disparate values that were unacceptable for population comparisons. Use of a single reference standard did little to improve comparability. Assay characteristics such as linearity, recovery, accuracy, and cross-reactivity to proinsulin and its primary conversion intermediates varied among the laboratories, and they did not readily explain differences in the measurements made from assay to assay. Use of the same assay kit in different laboratories did not always ensure comparable measurements. Linear regression of assay results from one laboratory to an arbitrarily chosen reference assay greatly improved comparability and demonstrated the potential value in comparing each assay to a reference method. The task force report defines acceptable assay characteristics and proposes a three-step process of insulin assay proficiency and comparability. A central reference assay and ongoing sample exchange will be needed to allow reliable comparisons of insulin measurements made in different laboratories. Rigorous quality control and continuous quality improvement are needed to maintain reliability of the insulin measurement.