It has been suggested that antibodies against non-acetylcholine receptor proteins of striated muscle are markers of the presence of a thymic epithelial tumor in patients with myasthenia gravis (MG). These antibodies may be measured using an immunofluorescence assay against striated muscle (anti-STR) or an ELISA with a recombinant 30-kd titin fragment (anti-MGT30). To directly compare anti-STR with anti-MGT30, we examined the sera of 276 consecutive patients with known or suspected MG. Definite diagnoses and thymic histology, if available, were correlated with the antibody assays. Of the 276 patients, 164 had MG. Thymic histology was obtained in 44 patients: 18 had lymphofollicular hyperplasia, 13 thymic epithelial tumors, 8 atrophy, and 5 were normal. When compared with anti-STR, anti-MGT30 showed a sensitivity of 69% (STR 77%), specificity of 100% (STR 56%, p = 0.026), negative predictive value of 82% (STR 77%), and positive predictive value of 100% (STR 56%, p = 0.003) for the identification of a thymic epithelial tumor versus thymic hyperplasia. We conclude that the anti-MGT30 ELISA is better than the anti-STR immunofluorescence assay for the diagnosis of paraneoplastic MG.