The prognostic performance of the 2 most commonly used staging systems for skeletal sarcoma (the American Joint Committee on Cancer [AJCC] and Musculoskeletal Tumor Society [MSTS] systems) have never been compared analytically. Another staging system originally proposed by Spanier has not yet been validated. Given the recent release of the 8th edition of the AJCC Cancer Staging Manual, this study was designed to directly compare these anatomic staging systems in a series of 153 high-grade, intramedullary osteosarcomas. Kaplan-Meier curves were plotted and pairwise comparisons between each stage category were performed. Predictive accuracy of each staging system for determining 5-year disease-free survival was evaluated by comparing areas under receiver-operating characteristic curves generated from logistic regression analysis. Multiple concordance indices were calculated using bootstrapping methods (200 replications). ρ2k and R2 were estimated as measures of the variation in survival outcomes explained by the regression models. The AJCC, MSTS, and a modified version of the Spanier staging systems showed similar discriminatory abilities and no significant differences in the levels of contrast between different tumor stages across staging systems. Addition of T-category information from each staging system contributed significant prognostic information compared with a Cox proportional hazard regression model consisting only of the presence or absence of metastatic disease as a measure of disease extent. Concordance indices and predictive accuracy for 5-year disease-free survival were not significantly different among the different staging systems either. Similar findings were observed after accounting for other important prognostic variables. Additional studies are necessary to determine performance parameters of each staging system for other types of skeletal sarcoma. Prognostic performance of osteosarcoma staging systems would also be improved by incorporating nonanatomic prognostic variables into staging algorithms.