The autoimmune origin of the Lambert-Eaton myasthenic syndrome (LEMS) was documented by passive transfer of its electrophysiological features from humans to mice with IgG. Freeze-fracture electron microscopy has demonstrated a loss of active-zone particles in human LEMS and in its mouse passive transfer model. These data imply that the active zones are targets of the pathogenic LEMS autoantibodies. Immunolocalization of the antibodies has been hindered, however, by a paucity of active-zone particles (about 50/micron2 normally and still lower in LEMS) and by diffusion artifacts in the immunoperoxidase method. To obviate these problems, we employed sensitive avidin-biotin detection systems, both peroxidase and ferritin labels, and quantitative immunoelectron microscopy and end-plate morphometry. We compared mice treated with LEMS IgG, control IgG, and no IgG. In all mice, nonspecific background staining was found in the basal lamina covering the muscle fibers and Schwann cells. When a single 10-mg dose of IgG was injected intravenously, IgG samples from 12 patients produced significant immunostaining of the mouse active zones; from 7 patients they did not. Higher doses of intraperitoneally injected IgG (20 mg, three times a day for 2 days, or 10 mg/day for 15 days) from each of 4 patients (3 of whose IgG previously transferred LEMS to mice) caused significant immunostaining of mouse active zones: (1) the mean density (no./micron presynaptic membrane length) of positive active zones was 0.91 in the immunoferritin study and 0.72 in the immunoperoxidase study (control values, 0.12 and 0.02); and (2) 43% of the ferritin particles in the primary cleft were concentrated at the active zones and the rest were scattered randomly (control value, 5.3%). The findings indicate that LEMS IgG binds to the active zones of the presynaptic membrane.