The natural history of breast cancer has been, and will be, extremely hard to study. Indirect evidence suggests that any cancer cell population may be phenotypically unique with multiple abnormal traits which decisively forge the clinical behavior. Histopathologic grouping will only give a very imperfect reflection of phenotypical individuality. It is proposed that essential determinants of the subsequent behavior have already been fixed during the induction period, i.e. long before the clinical debut. This may to some extent be reflected by the DNA content of the tumor cells. Metastases to axillary lymph nodes is strongly determined by size of primary tumor according to a function which suggests a smooth linear risk for lymphatic spread over a broad size interval. It is suggested that each cell of any given cancer may be endowed with a small chance of dissemination to axillary lymph nodes which may be essentially constant during the entire natural life history. Number of cancer cells of the primary tumor would then be the decisive determinant for risk of metastatic dissemination. It is not necessary to invoke qualitative progression from 'non-metastasibility' to 'metastasibility'. Small differences exist in the natural history of different histological types of breast cancer.