In prognosis of breast cancer different parameters are in current use. Along with clinical staging the most important parameter appears to be histologic grading. Features of the grading such as nuclear pleomorphism proved to correlate closely with the proliferative activity and aggressiveness of the tumors. Because of difficulties in assessing and classifying the degree of nuclear pleomorphism by usual microscopy, the authors applied methods of digital image analysis. The study is a retrospective analysis of paraffine slides from the primary lesions of 60 breast cancers with 10 to 16 years of follow-up evaluation. Using large sets of different parameters defining nuclear morphology and chromatin structure the authors extracted criteria with prognostic importance. These included nuclear area in micron 2, eccentricity, integral optical density per micron 2, average area of a chromatin region, integral optical density of a chromatin region per micron 2, and the number of central chromatin regions per micron 2. The results demonstrate that the criteria used enable prediction of prognosis with an accuracy of 92%.