To predict clinical outcome, we studied 42 paragangliomas from 37 patients by routine histology, immunohistochemistry, and electron microscopy. A panel of antisera to neuron-specific enolase (NSE), chromogranin, and met-enkephalin was used to identify chief (type I) cells, and S-100 protein and glial fibrillary acid protein (GFAP) sustentacular (type II) cells. The intensity of staining of type I cells and the density of type II cells were assessed semiquantitatively (0 to 4+) in a total of 38 tumors. A total of 23 of 24 low-grade tumors (solitary, multiple, or associated with other neoplasms; 95.8%) contained type II cells immunoreactive with either S-100 protein or GFAP, and all were positive when S-100 protein and GFAP were used in combination. Five of the nine intermediate-grade (recurrent and/or locally aggressive) tumors were identified as glomus jugulare tumors (GJT). Three intermediate-grade GJTs were devoid of GFAP-reactive type II cells and four GJTs were negative for S-100 protein. Type II cells were identified in only one of five high-grade (malignant) paragangliomas and that tumor contained vanishingly rare cells that were weakly S-100 protein positive but GFAP negative. Sustentacular cell density and chief cell staining intensity were both inversely related to tumor grade. The most sensitive chief cell marker was NSE (92.1%), followed by chromogranin (84.2%). The least sensitive (73.0%) and specific marker was met-enkephalin. Combinations of NSE or chromogranin with met-enkephalin identified chief cells in all cases. Electron microscopy identified neurosecretory granule-containing chief cells, but was of less value in delineating sustentacular cells because of their scarcity and the absence of specific features. By comparison, immunohistochemistry was superior in identifying sustentacular cells. The use of an immunohistochemical panel, in addition to routine histology, can confirm the diagnosis of a paraganglioma and can give an indication of the likely prognosis for a patient.