The purpose of this study was to describe one method of classification, based on anatomic-mammographic correlations, developed by Tabar. We also wanted to examine how the mammograms categorized as low- and high-risk according to Tabar and Wolfe criteria related to each other and to three selected risk factors for breast cancer. The study materials are based on questionnaires and mammograms from 3.640 Norwegian women, aged 40-56 years, participating in the third Tromso study. The mammograms were categorized into five groups. Line drawings and their pathologic correlates of the five patterns are described in detail. The Tabar classification is based on anatomic-mammographic correlations, following three-dimensional (thick slice technique) histopathologic-mammographic comparisons, rather than simple pattern reading (Wolfe classification). For analysis patterns I-III (Tabar) and N1 and P1 (Wolfe) were grouped into low-risk groups and patterns IV and V (Tabar) and P2 and DY (Wolfe) into high-risk groups. The overall agreement on high-risk versus low risk for the two classifications was 54% with a κ-value of 0.22. The study displays that the strength of association between high-risk mammographic patterns and the three selected risk factors parity, number of children and age at first birth is of greater magnitude when the Tabar instead of the Wolfe classification is applied. More patients are needed to compare the classification directly with the risk of cancer. This study indicates that further development of the classification of mammograms may increase the usefulness of mammographic patterns in research and clinical practice.