Microarray analysis assessed expression of approximately 12,000 genes in 50 high-grade gliomas. A two-class prediction model, based on a subset of glioblastomas and anaplastic oligodendrogliomas with classic histology, predicted the classification of clinically common, histologically nonclassic samples. This model provided a more accurate predictor of prognosis in nonclassic lesions than did pathological classification. Left, diagrammatic representation of the top 50 features for classic glioblastoma and anaplastic oligodendroglioma, with red indicating high expression, and blue indicating low expression, relative to the mean. Right, survival curves of patients with nonclassic glioblastomas (dashed line) and nonclassic anaplastic oligodendrogliomas (solid line). For details, see the article by Nutt et al. on page 1602 of this issue.