For automated astrocytoma grading morphometric parameters are determined by means of an image analysis system and a special Ki-67(MIB1)/Feulgen-staining method allowing the quantification of the essential characteristics of malignant gliomas: growth pattern, cellularity, proliferation index and nucleus pleomorphism. Based upon a cluster analytical approach a grading scale resembling the WHO-scheme is established which is suitable for automatic glioma grading purposes (HOM-scale). For automatic glioma grading backpropagation neural networks are employed. The results are compared with those of a classical multivariate discriminant classificatory analysis. The presented approach can also be employed for automatic grading of other tumour entities.