On type 1 newly diagnosed and on insulin treated diabetic patients, anti-insulin autoantibodies (IAA) and antibodies (IA) having the same specificity are respectively induced. Such immune response may be evaluated either by radiobinding assay (RBA) or enzyme-linked immunosorbent assay (ELISA). Both methodologies have been compared at previous International Workshops, which pointed out discrepancies in results. In this work, IAA/IA prevalence was assessed by displacement RBA and ELISA, in normal subjects, type 2 (treated with hypoglycaemic agents), insulin treated and newly diagnosed type 1 diabetic patients. Results showed a lack of RBA-ELISA agreement. An attempt was then made to determine whether such results were, at least in part, attributable to iodination site in Tyr-A14. For this purpose parallel RBA assays were carried out by using radiolabelled insulin at A14 and A19 Tyr residues. Control sera and samples from insulin treated and type 1 newly diagnosed diabetic patients were tested. Our results suggest that labelling position is not involved in artifactual binding of tracers, at least as a systematic phenomenon. In the majority of cases the variability in RBA-ELISA signal ratios are best explained in terms of differences in the basic principles operating in both methods instead of artifacts due to tracer preparation.