The goal of screening tests for breast cancer is early detection and treatment with a consequent reduction in mortality caused by the disease. Screening tests, however, might produce misleading diagnoses and potentially significant emotional, financial and health costs. The effectiveness of a breast screening program has effects on the quality of life of the target population. Even if the screening units regularly attain coverage targets, it remains essential to ensure that women receive the same high standard of service wherever they live. In order to assess the relative efficiency of individual screening units we use stochastic D.E.A. models, which can be used as reliable tools for external audit. The technique is tested on breast cancer screening data of two Italian regions.