The aims of this study were to investigate, in patients with newly diagnosed small-cell lung carcinoma (SCLC), whether or not there may be a relationship between the presence, type or titer of circulating neuronal autoantibodies and (i) the extent of SCLC dissemination at presentation, (ii) the development of peripheral neuropathy during platinum chemotherapy, (iii) survival time. We studied stored serum from 58 patients with uncomplicated SCLC who had participated in two trials conducted by the North Central Cancer Treatment Group (NCCTG); 29 had extensive disease and 29 had limited disease. No patient had neuropathy or other neurological or paraneoplastic problems at the time of enrollment but each group included 14 or 15 patients respectively who developed peripheral neuropathy in the course of chemotherapy. We tested five consecutive serum specimens from each patient in blinded fashion by (i) an indirect immunofluorescence assay optimized to detect neuron-restricted nuclear and cytoplasmic antibodies (triple substrate of mouse cerebellum, gut and kidney), and (ii) immunoprecipitation assays to detect neuronal Ca2+-channel-binding antibodies (N-type and P/Q-type). Sera that were positive by immunofluorescence were analyzed further by Western blotting. Neuronal autoantibodies were significantly more frequent in patients who had limited SCLC at presentation (12/29 or 41% positive) than in those with extensive SCLC (5/29 or 17% positive, P = 0.02). Neuronal autoantibodies of nuclear or cytoplasmic specificity were found in 50% of the seropositive patients with limited SCLC (21% of the total group), but in no patient with extensive SCLC (P = 0.01). The frequency of neuronal autoantibodies did not differ significantly among patients who did and did not develop peripheral neuropathy. Titers fell progressively during chemotherapy and did not rise again when peripheral neuropathy became clinically evident. This argues against a synergism between drug toxicity and neuronal autoimmunity as the mechanism of platinum-associated peripheral neuropathy. Seropositivity for neuronal autoantibodies did not affect the survival of patients with either limited or extensive SCLC. It is conceivable that the immunosuppression attendant on combined cisplatin/etoposide therapy cancels a pre-existing protective antitumor immune response (presumably cytotoxic-T-cell-mediated) for which the nuclear and cytoplasmic paraneoplastic IgG autoantibodies serve as a surrogate marker. Testing of this hypothesis would require the survival of seropositive and seronegative patients to be compared in a larger trial, using a therapeutic modality that does not compromise immunocompetence.