Purpose To determine whether the clinical behavior of giant cell lesions (GCLs) or their anatomic location can be differentiated by histologic criteria alone. Materials and Methods We performed a retrospective study of patients with GCLs treated at Massachusetts General Hospital between 1993 and 2008. Predictor variables were histologic parameters: number of giant cells (GCs) per high-power field, number of nuclei per GC, GC size, stromal cellularity, stromal type, presence of hemorrhage and reactive osteoid, and blinded pathologists' prediction of location and behavior. Outcome variables were clinical behavior (aggressive or nonaggressive) and GCL location, that is, maxillofacial (MF) or axial/appendicular (AA). Descriptive and bivariate statistics were computed with statistical significance set at P ≤ .05. Results The sample included 88 subjects: 41 MF GCLs (35 aggressive) and 47 AA GCLs (28 aggressive). Aggressive AA lesions had more GCs per high-power field, larger mean GC size, and increased stromal cellularity, and they more frequently had a mononuclear stroma when compared with aggressive MF lesions ( P < .05). There were no significant histologic differences between aggressive and nonaggressive MF lesions or between nonaggressive MF and nonaggressive AA lesions. Aggressive AA lesions had more nuclei/GC than nonaggressive AA lesions ( P = .03). Using histologic criteria only, blinded pathologists predicted clinical behavior in only 45% of cases (κ = 0.19, P = .09). They predicted a lesion's location in 82% of cases with fair agreement (κ = 0.44, P < .01). Conclusions Results of this study indicate that histologic differences between aggressive and nonaggressive GCLs are insufficient for pathologists to differentiate them consistently regardless of location.