The main advantage of early diagnosis of cancer is the reduction of tumor size at initial treatment and thereby an increase in the proportion of patients without distant dissemination. This benefit is illustrated by the example of breast cancer. A model of its natural history was built using data extracted from the files of over 4000 patients followed in the same institution for 20 to 35 years. The model was used to quantify the impact of tumor size, histologic grade, and lymph node involvement on the probability of distant spread. The relationships were found to be highly significant. The model also and unexpectedly revealed that the tumors progress while they grow; avoiding histologic progression is therefore another advantage of early diagnosis. The model showed that residual tumor can be a nidus for distant dissemination, and that consistency between the prediction of the model and the results of post-operative radiotherapy is satisfactory. Conversely to what is often stated, the benefits of post-operative radiotherapy appear to be greater for small tumors, even in the absence of lymph node involvement, than for large ones. The model could be used to help improve screening strategies, but more data are required, in particular for the young age range.