Schwannomas are benign peripheral nerve sheath tumors believed to be composed purely of cells with ultrastructural features of Schwann cells; these tumors are believed to develop eccentrically from the surface of nerves and not to contain axons, other than immediately beneath the capsule. This concept has recently been disputed in cases associated with neurofibromatosis type 2. The usual presence of intratumoral axons in neurofibromas is said to allow easy distinction from schwannomas. Eighty sporadic schwannomas (20 conventional, 20 cellular, 20 ancient, 10 gastric, and 10 plexiform) were retrieved from the authors' files. Hematoxylin-and-eosin stained slides were reviewed, diagnoses were confirmed and all tumors were stained for S-100 protein and neurofilament protein (NFP). The amount (rare, focal, multifocal, and diffuse) and distribution (central and/or peripheral) of axons within the tumors were analyzed. All tumors were strongly and diffusely positive for S-100 protein (nuclear and cytoplasmic staining). NFP-positive axons were identified in 11 of 20 (55%) conventional schwannomas (2 rare, 4 focal, 3 multifocal, and 2 diffuse; 5 central, 4 peripheral, and 2 central and peripheral) and in 15 of 20 (75%) cellular schwannomas (3 rare, 6 focal, and 6 multifocal; 12 central, 1 peripheral, and 2 central and peripheral). Of the 20 ancient schwannomas, 7 cases (35%) showed intratumoral axons, highlighted by NFP immunostaining (1 rare, 4 focal, 1 multifocal, and 1 diffuse; 4 peripheral, 2 central, and 1 central and peripheral). Most cases of gastric schwannoma showed no evidence of intratumoral axons; 9 cases (90%) were negative for NFP and only 1 case (10%) was positive (focal and central). Seven of 10 cases (70%) of plexiform schwannomas were negative for NFP, whereas only 3 cases (30%) showed positive axons (2 multifocal and 1 focal; 3 central). The unexpected but quite frequent presence of intratumoral axons in schwannomas argues against conventional views of these lesions' pathogenesis as an eccentric encapsulated lesion and raises the possibility that a more diverse cell population, perhaps more closely resembling neurofibromas, may constitute these neoplasms. Although NFP-positive axons were most often present in the conventional and cellular variants of schwannoma, their presence was also observed in a minority of ancient, gastric and plexiform schwannomas. Differentiation between neurofibroma and schwannoma in cases with overlapping cytoarchitectural features should not be based solely on the presence or absence of NFP-positive axons within a given tumor.