Updated guidelines on breast cancer screening have been published by several major organisations over the past five years. Recommendations vary regarding both age range, screening interval, and even on whether breast screening should be offered at all. The variation between recommendations reflects substantial differences in estimates of the major benefit (breast cancer mortality reduction) and the major harm (overdiagnosis). Estimates vary considerably among randomised trials, as well as observational studies: from no benefit to large reductions, and from no overdiagnosis to substantial levels. The estimates vary according to the methodology of the randomised trials, and the design of the observational studies. Guideline recommendations reflect the choice of evidence informing them. While there are well-developed tools to deal with randomised trials in guideline work, these are not always used, or they may not be followed as recommended. Further, results of trials performed decades ago may no longer be applicable. For observational studies, the framework for inclusion in guidelines is not similarly well-developed and there are methodological concerns specific to screening interventions, such as small effects in absolute terms. There is a need for agreement on a hierarchy of observational study designs to quantify the major benefit and harm of cancer screening. This review provides a summary of recent guidelines on breast cancer screening and their major strengths and weaknesses, as well as a short overview of the major strengths and limitations of observational study designs. There is a need for agreement on a hierarchy of observational study designs in this field.