Direct assays for PAIgG [platelet-associated IgG] are useful in diagnosing ITP [idiopathic thrombocytopenic purpura] and investigating mechanisms of thrombocytopenia. A serum assay for S-PBIgG [serum platelet-bindable IgG] would offer technical advantages in transportation and storage of samples and could prove useful in the diagnosis of thrombocytopenia caused by alloantibodies. The development of an assay for S-PBIgG that is positive in the majority of patients with immune thrombocytopenia is reported. This was accomplished by empirically testing a number of variables, including the anticoagulant used to prepare target platelets, the presence of a fixative, incubation time and the target platelet count. The development of this assay was facilitated by using a microtiter modification of the antiglobulin consumption assay, which was validated by simultaneously determining PAIgG with each method (r = 0.91, P < 0.01, n = 42). The serum assay was then applied to the study of several types of immune thrombocytopenia. PAIgG and S-PBIgG were simultaneously quantitated on platelets and sera obtained from 42 patients with idiopathic thrombocytopenia. PAIgG was elevated in 38; S-PBIgG was elevated in 34. Eleven patients had a normal or only moderately elevated S-PBIgG but a considerably elevated PAIgG. In these patients the unbound platelet antibody or immune complex may have platelet specificity. In the remaining patients there was a close correlation between the level of PAIgG and S-PBIgG (r = 0.81, n = 31) indicating that the bound platelet antibody or immune complex is in dynamic equilibrium with the unbound. The potential application of the assay for S-PBIgG was demonstrated in 3 leukemic patients who were refractory to platelet transfusions and in 2 mothers who had infants with alloimmune neonatal thrombocytopenia. In these patients elevated S-PBIgG was associated with a normal or only slightly elevated PAIgG.