The discovery of islet cell antibodies (ICAs) was the prelude to the understanding that type 1 diabetes mellitus (T1DM) is a chronic autoimmune disease. The issue regarding whether or not the measurement of ICAs should be completely replaced by biochemical markers detecting islet autoantibodies (AAs) for the prediction of T1DM has been the subject of endless international debates. In light of this controversy, we assessed the current role of ICAs as a predictive marker for T1DM progression. We examined a cohort of 1484 first-degree relatives (FDRs) of T1DM probands from the Children's Hospital of Pittsburgh Registry. These relatives were consecutively enrolled between 1979 through 1984 and followed up to 22 yr. Serum obtained at the time of enrollment was assayed for ICAs, glutamic acid decarboxylase (GAD)65, insulin A (IA)-2 AA, and insulin AAs (IAAs). In FDRs who had ICAs in addition to GAD65 and IA-2 AAs, the cumulative risk of developing insulin-requiring diabetes was 80% at 6.7 yr of follow-up, whereas this risk in those with GAD65 and IA-2 AAs without ICAs was only 14% at 10 yr of follow-up (log rank: P < 0.00001). Cox regression analysis showed that diabetes risk was significantly associated with the presence of ICAs in both subjects with low titer and high titer GAD65 and IA-2 AAs. The addition of IAAs in GAD65 and IA-2 AA-positive relatives did not increase the cumulative risk for conversion to insulin-treated diabetes. We provide evidence that a subgroup of ICAs predicts a more rapid progression to insulin-requiring diabetes in GAD65 and IA-2 AA-positive relatives and should remain part of the assessment of T1DM risk for intervention trials. In addition, these findings provide impetus for efforts to identify a novel islet autoantigen(s) reactive with this ICA subset.