Diffusion tensor imaging (DTI) has the potential to improve prostate cancer detection, since anisotropy is expected to correlate with tumor aggressiveness and differentiation. Differences in fractional anisotropy between cancer and normal tissue have been observed, although data is somewhat contradictory. A problem with DTI is its inability to distinguish low anisotropy from high orientation dispersion. In this study, we map the anisotropy independent of orientation in the prostate, by the use of a novel diffusion-encoding technique that permits encodings with variable b-tensor shapes. The microscopic anisotropy was found to be generally higher in cancer than in normal prostatic tissue.