Studies of preneoplasias and carcinomas of the mammary gland have been conducted by means of automated microscopic image analysis, on the basis of previous results as well as with references to the international literature, for 2 purposes: 1. Determination in the context of a clinical follow-up study of the individual carcinoma risk for patients with proliferative fibrocystic breast disease (mastopathy) and 2. Preparation of an objective automated grading of ductal breast carcinomas for better assessment of prognosis. Against the background of the assumption that the majority of carcinomas and precancerous lesions of the breast originate from the terminal ductal lobular unit, an effort is made to determine, independent severity of the mastopathy, the biological valence of solid, cribriform, and papillary ductal epithelial proliferations (no, possible or inevitable preneoplasia). Proliferation patterns without and with atypical features are checked for their similarity with intraductal and invasive carcinomas (similarity principle) and are additionally examined for differences, depending on localization and distance from tumour (topological principle). Automated histological tumour grading is to distinguish with greater subtlety within the large heterogeneous group of moderately differentiated carcinomas and is to objectify and thus facilitate the difficult task of delimitation of moderately and poorly differentiated carcinomas. Nuclear grading will be the major basis for classification along these lines. Objectified and reproducible tumour grading is believed to be extremely helpful in better prognostication of breast cancer.