This paper reports a retrospective study of 69 breast biopsies carried out for surgically occult but radiologically apparent lesions. Specimen radiography was required to confirm excision. The group included 17 carcinomas. Review of the pre-operative mammogram reports provided by a 'pool' of general radiologists demonstrated a high sensitivity (88%), but poor specificity (32%), with an overall accuracy of 46%. 'Blind' review of the mammograms by one experienced mammographic radiologist showed high sensitivity (100%), good specificity (73%) and overall accuracy of almost 80%. This result shows the need for the most experienced radiologist available to be involved in deciding which of these difficult lesions require biopsy. This will reduce unnecessary breast surgery and highlights the role of clinical and radiological review instead of biopsy, the need for continuing self audit by radiologists and the need for regular communication between clinicians, radiologists and pathologists.