Radiation dose and the associated risk to the breast are being studied in much detail with the aim of justifying breast cancer screening. The mean glandular dose has been proposed as the proper quantity to describe risks incurred by radiation. X. Wu (US) and D. Dance (UK) were the first authors to standardize breast dosimetry and their conversion tables have been used throughout. Doses can be calculated for specific population samples or for individual patients, in which case an accurate estimate of the glandular tissue fraction should be known. From specific DICOM tags in the digital images and access to tube output measurements from the medical physics QA, patient dose assessment can be largely automated. Other (traditional) approaches mimic clinical exposures on a Perspex phantom and use these exposures to estimate the breast dose to a typical breast. Any local data, both patient dose data and doses to Perspex, can these days be compared to an extensive amount of published dose values. Radiation dose and image quality should both be considered in screening applications. Better image quality can have a significant effect on diagnostic performance. As only a small increase in performance (ex. detected fraction) gained can mean an enormous amount of women saved, image quality settings should be optimized and not minimized. Risk benefit investigation should therefore focus on performance rather than dose. For a few screening programs, the benefits could be proven. Emerging technologies should be investigated along the same basic principles.