We compared quantitative analysis of conventional MR and magnetization transfer (MT) images with visual morphologic assessment of standard images in the classification of gliomas. Thirty-two patients with gliomas were imaged preoperatively. Relative signal intensities on T2- and T1-weighted images, Gd-enhancement, signal heterogeneity on T2-weighted and Gd-enhanced T1-weighted images and MT ratios were measured. In 16 astrocytomas, the MT ratios were correlated with the volume fraction of tumor cell nuclei in histological sections. Among the conventional sequences T2-weighted images were most accurate for quantitative classification; signal heterogeneity was more accurate than relative signal intensity. MT ratios were superior to quantitative analysis of conventional images and equal to visual morphologic evaluation in discriminating between low-grade and high-grade gliomas and correlated with the volume fraction of nuclei in the tumor tissue (r = 0.71, p < .01). The results indicate that quantitative analysis is generally of minor value in the grading of gliomas. In spite of sensitivity to tissue changes associated with malignancy MT imaging cannot significantly improve MR classification of gliomas.