We found a high prevalence of pulmonary and extrapulmonary neoplasms in patients with pulmonary eosinophilic granuloma (PEG) who were studied at our institution. Among 21 patients with PEG, 10 (48%) had associated benign (one patient) or malignant (nine patients) tumors. Patients with tumors were older at the time of diagnosis of PEG (48.9 vs 34.5 years). Tumors included three lung carcinomas, one pulmonary carcinoid tumor, two lymphomas, five extrapulmonary carcinomas, and one mediastinal ganglioneuroma. Two malignant neoplasms developed in each of two patients. Six tumors preceded, three followed, and three occurred concomitantly with the diagnosis of PEG. Slides from eight PEG-associated tumors and 18 control neoplasms from patients without PEG were also stained immunohistochemically for S100 protein. Four PEG-associated (50%) and 11 control (61%) tumors contained S100 protein-positive interstitial cells. Our study suggests, but does not prove, that there may be more than a random association between PEG and neoplasms. Cigarette smoking, moreover, is an important risk factor for both PEG and lung carcinomas. Our immunohistochemical findings indicate that S100 protein-positive cells in tumors usually bear little or no relationship to PEG. In patients with an underlying malignant neoplasm, PEG simulates pulmonary metastases clinically and, occasionally, histopathologically.