Plasma insulin concentrations of insulin-treated diabetic patients must be measured after removal of insulin antibodies, usually by precipitation with polyethylene glycol (PEG). Details of the procedure vary between laboratories; commonly, frozen plasma is thawed and incubated at 37 degrees C to restore a presumed equilibrium between free and antibody bound insulin before extraction. The present study was designed to investigate methodological factors that could affect the measured free insulin concentration. In normal subjects PEG extraction of globulins did not disturb measurement of insulin concentrations, whether carried out after incubation for 2 h at 37 degrees C, or storage at -20 degrees C, in either order. Freezing or incubation of PEG extracts of plasma from insulin-treated patients also failed to disturb the measured concentrations of free insulin. When plasma from patients was incubated for 2 h after storage, a marked scatter (51-272%) of measured results occurred when compared to bedside extraction. This problem was not overcome by buffering with HEPES or storage at a lower temperature (-40 degrees C). Incubation at 0 degrees C also severely disturbed the apparent concentrations. Incubation of plasma before extraction and freezing also disturbed the measured result, a problem not corrected by maintaining near physiological pH. Total insulin concentrations measured on acid extracts were not disturbed by any of these manoeuvres. The temperature of centrifugation of blood at the time of venepuncture did not influence the result.(ABSTRACT TRUNCATED AT 250 WORDS)