The contribution of glutamate decarboxylase (Mr 65000) antibodies to the reactivity of islet cell cytoplasmic antibodies with the 'whole' islet staining pattern from patients with newly diagnosed Type I diabetes was investigated. Diluted sera (n = 10) were preincubated with increasing concentrations of purified recombinant human islet glutamate decarboxylase (Mr 65000) and the change in islet cell cytoplasmic antibody binding was evaluated by quantitative immunocytochemistry. Binding to islet cells was partially blocked by glutamate decarboxylase in 9/10 diluted sera; the maximum blocking obtained at high concentrations of glutamate decarboxylase (5 micrograms/ml) was 36% (median, range 24-61%). In contrast, binding to islet cells in three diluted sera (two polyendocrine patients without Type I diabetes and one patient with newly diagnosed Type I diabetes) with the 'selective' islet staining pattern was totally blocked by glutamate decarboxylase. The concentration of glutamate decarboxylase required to achieve maximum blocking was less for the 'whole' islet (0.4-8.0 micrograms/microliters undiluted serum) compared to the 'selective' islet (20-645 micrograms/microliters undiluted serum) positive sera. All sera were positive for glutamate decarboxylase antibodies in an immunoprecipitation assay using 35S-methionine labelled extract of baby hamster kidney cells transfected with glutamate decarboxylase. However, the binding activity of these antibodies was less in the sera positive for the 'whole' islet compared to the 'selective' islet staining pattern. In conclusion, glutamate decarboxylase antibodies contribute partially to the reactivity of islet cell cytoplasmic antibodies of the 'whole' islet staining pattern in the sera of newly diagnosed patients with Type I diabetes, and totally to reactivity of the 'selective' islet staining pattern. The antigens recognized by the other antibodies contributing to the 'whole' islet reactivity remain to be defined.