The effects of plasmapheresis on islet autoantibody levels, C-peptide (beta-cell function), and hemoglobin-A1c (HbA1c, metabolic control) were tested in a prospective blinded study of 18 newly diagnosed insulin-dependent diabetes mellitus (IDDM) patients randomly assigned to receive plasmapheresis (P), carried out as double filtration, or sham (S) treatment at diagnosis and 3 months thereafter. At diagnosis, 6 of 8 patients (75%) in group P and 9 of 10 patients (90%) in group S had islet cell antibodies (ICA), whereas 4 of 8 (50%) and 7 of 10 (70%) patients, respectively, had glutamic acid decarboxylase antibodies (GAD65-Ab), with no significant differences between the groups in ICA and GAD65-Ab levels. After 6 months, P patients showed significantly lower ICA levels than S patients (11 +/- 6 and 128 +/- 47 Juvenile Diabetes Foundation International Units, respectively; P < 0.02) due to an increase in ICA levels in 8 of 9 (88%) of the S patients not seen in P patients (P < 0.002). Concurrently, HbA1c stabilized in P, but not in S, patients and was significantly lower by 24 months (6.58 +/- 0.54% vs. 9.76 +/- 1.21%; P < 0.05). Moreover, fasting C-peptide increased significantly (214 +/- 11 pmol/L; P < 0.05) over the first 6 months in P. After the initial 6 months, ICA levels tended to decrease in all patients and were not detected after 60 months. GAD65-Ab levels were not influenced by plasmapheresis and, also in contrast to ICA, increased significantly (P < 0.05) in the whole study population after 60 months. In fact, 4 initially negative patients became GAD65-Ab positive after diagnosis (in 2 patients > 24 months after diagnosis). We conclude that plasmapheresis of newly diagnosed IDDM patients does not change subsequent GAD65-Ab levels, but ICA are significantly decreased with associated improved C-peptide and HbA1c levels.