We studied 217 consecutive tumors of bone by flow cytometric analysis of nuclear DNA concentration after staining with propidium iodide. A diagnosis and histological grade (benign, low-grade, or high-grade sarcoma) were assigned to each tumor on the basis of staging data (with the exception of the forty-six giant-cell tumors, which, although indistinguishable histologically, were divided according to the flow cytometric pattern into two distinct groups), and we quantitatively studied the flow cytometry data to assess the percentages of cells in diploidy, tetraploidy, or aneuploidy. When compared, the mean values for the flow cytometric data for the three grades showed significant differences. Criteria were established for the three classes of tumors: for benign tumors, less than 11 per cent tetraploidy and no aneuploidy; for low-grade sarcomas, more than 11 per cent and less than 17 per cent tetraploidy, and no aneuploidy; and for high-grade tumors, either more than 17 per cent tetraploidy or aneuploidy. Tests for compliance for all groups of tumors (excluding the forty-six giant-cell tumors)--benign, low grade, or high grade--were significant for most of the benign lesions (with the exception of chondroblastoma and fibrous dysplasia) and for the high-grade sarcomas (with the exception of round-cell tumors). The low-grade sarcomas did far less well, based principally on the failure of the low-grade chondrosarcomas, chordomas, and adamantinomas to comply with the criteria. An attempt to assess the value of the system as a predictor of metastases showed that a low percentage of diploid cells (less than 75 per cent) and the presence of an aneuploid peak correlated statistically with the development of metastatic disease, but the usefulness of this observation could not be fully assessed because of multiple variables, associated principally with treatment.