Plasma levels of islet cell cytoplasmic and cytotoxic antibodies were determined in 10 children with insulin-dependent diabetes mellitus (IDDM) treated with plasmapheresis shortly after diagnosis, and in 9 children with IDDM treated by conventional means alone. Islet cell cytoplasmic antibody (ICA) titers were determined by indirect immunofluorescence using unfixed sections of human pancreas, and islet cell cytotoxic antibody levels were determined in a complement-dependent antibody-mediated cytotoxicity (C'AMC) assay using a human fetal cloned insulin-producing cell line (JHPI-1) as target. Before plasmapheresis, ICA was present in 7 out of 10 children and C'AMC was positive in 4. Four successive treatments with plasmapheresis did not consistently decrease plasma levels of ICA or C'AMC. ICA was present in 15 out of the total 19 children at diagnosis, and titers of ICA decreased in 12 out of 15 subjects by at least 1 degree of dilution (1:3) at 18-30 months follow-up, whether or not they had been treated with plasmapheresis; C'AMC was positive in 6 out of the 18 children at diagnosis and decreased in 2 out of 6. Plasma levels of C-peptide did not differ at diagnosis but remained higher in the plasmapheresis treated diabetic children at 3 and 18-30 months follow-up. Neither ICA titers nor C'AMC levels correlated with plasma C-peptide responses at 18-30 months. It is concluded that plasmapheresis decreases ICA and C'AMC but is followed rapidly by a rebound effect, and does not affect the rates at which these islet cell antibodies decrease with increasing duration of IDDM.