Numerous trials have shown, that breast cancer have highly variable rates of growth. It is assumed, that the rates (relative growth rates) decelerate with increasing tumour size. The Universities of Heidelberg and Louisville carried out a retrospective statistical analysis of the mammographically measured growth rates of 448 screening patients until breast cancer diagnosis. The analysis did not include fast-growing carcinomas appearing between mammograms for which only one mammogram was available or some cancer, where growth was not detectable by mammography. Generalized logistic curves provided the best fit to the data on the increase in tumour size, as observed in mammograms. Large variations in individual tumour doubling times were found, from extremely fast-growing to extremely slow-growing tumours. The results are relevant for patient prognosis, for the evaluation of therapy, and for screening strategies.