The main purpose of this in vivo study was to provide a detailed description of the T1- and T2-relaxation processes in intracranial tumors at 1.5 T. A total of 100 patients were investigated. Optimal experimental conditions were carefully observed, including the use of long TR values. T1 determination was based on a partial saturation inversion recovery sequence covering 12 or 6 data points. T2 determination involved a multiple spin echo sequence with 32 echoes. Calculations included biexponential analysis of the 12-point T1 data and all T2 data obtained. The results were evaluated in accordance with histopathology. The T1- and the T2-relaxation times of the prevailing tumor types were significantly different (p < 0.0005). However, biologic scatter and overlap between tumor types were considerable. In particular, no discrimination between benign and malignant tumor growth was possible. Biexponential evaluation did not increase the specificity, although a biexponential relaxation behavior was recognized in 37% of the T2 curves. The results indicate that tissue heterogeneity is responsible for most of the scatter in the relaxation times. It is concluded that tissue characterization by MR imaging, based solely on relaxation time measurements, seems to be of no value in the differentiation of intracranial tumors.