To develop a cost-effective approach to the detection of clinically significant lesions in grossly benign breast biopsies, we reviewed 384 consecutive breast biopsies performed because of a palpable mass in which gross examination revealed only benign fibrofatty tissue. All tissue was submitted for histologic examination (total of 3,342 blocks). Carcinoma or atypical hyperplasia (AH) was detected in 26 cases (6.8%). This group included 12 cases of lobular carcinoma in situ, four cases of atypical lobular hyperplasia, four cases of ductal carcinoma in situ, three cases of atypical ductal hyperplasia, and three cases of invasive carcinoma. In 25 of the 26 cases, at least a portion of the lesion was present in fibrous parenchyma; in only one case of atypical lobular hyperplasia was the lesion present exclusively in grossly fatty tissue. If, as a maximum, the first five tissue blocks had been submitted for each case and the remaining tissue was submitted only for cases of carcinoma or atypical hyperplasia, 1,386 (41%) fewer blocks would have been submitted. However, six (23%) of the carcinomas or atypical hyperplasias would have gone undetected. In contrast, submitting the first 10 blocks in each case would have resulted in 610 (18%) fewer blocks submitted but in the detection of all cases of carcinoma and atypical hyperplasia, except for a single focus of lobular carcinoma in situ in one case. We also used statistical models to investigate further the factors important in determining the probability of detecting carcinoma or atypical hyperplasia in grossly benign breast biopsies. We found that the percentage of blocks containing the lesion and the actual number of blocks submitted were the most important of these factors and that the total specimen size was a relatively minor determinant of the probability of detection. We conclude that (a) the likelihood of detecting carcinoma or atypical hyperplasia exclusively in the adipose tissue component of grossly benign breast biopsies is extremely low, and (b) a possible cost-effective method of sampling grossly benign breast biopsies consists of initially submitting a maximum of 10 blocks of fibrous parenchyma for each case, then examining the remaining tissue histologically only if carcinoma or atypical hyperplasia is found among these blocks.