A recent international workshop documented marked interlaboratory variation in end point titers of standard islet cell antibody (ICA) positive sera. End titers were lower using a modified assay which utilizes fluorescein labeled protein A (ICA-pA) rather than fluoresceinated anti-IgG (ICA-IgG) to detect antibody binding to islets. In this study we sought to compare directly two ICA assays with respect to future development of IDDM. Sera were obtained from 26 prospectively evaluated high risk subjects identified by family screening or history of transient hyperglycemia and 12 normal controls. As expected, end point titers for ICA positive sera were 10 times greater using the ICA-IgG assay than with the ICA-pA assay. However, despite higher end point titers, the ICA-IgG assay failed to detect more 'prediabetics' and showed a prozone effect. Fourteen subjects were positive at a 1:2 dilution using the ICA-pA assay. Only 10 of these 14 were positive at a 1:2 dilution using the ICA-IgG assay but all became positive at greater sera dilutions. No normal controls were positive using either assay. A similar prozone was observed with anti-islet monoclonal antibodies A2B5 and 4F2. Sera from 14 long-standing IDDM patients (where titers of ICA may have decreased relative to time of onset of diabetes) which were negative using ICA-pA were also assayed using ICA-IgG. Five sera positive for ICA-IgG but negative for ICA-pA were identified. In addition two sera in which a prozone effect was seen with ICA-pA were identified.(ABSTRACT TRUNCATED AT 250 WORDS)