Microscopic foci of multicentric cancer were detected in 121 of 904 breasts surgically removed for a clinically overt, invasive cancer. This incidence of 13.4% is regarded as a conservative estimate since examples of such lesions occurring in the same quadrant as the dominant mass, except in those instances in which the latter was present within the tail of the breast or beneath the nipple, were excluded from the analysis. Further, this data was obtained from only one randomly selected block of the quadrants, and in 41% of the cases only one or two were available for study. Multiple multicentric cancers were found in the same breast in two and three quadrants in 11.6% and 5.8% of the cases respectively. I n 9.3% of the cases the multicentric cancers were designated as noninvasive (lobular in situ and/or intraductal) and in 4.1% invasive. An attempt to correlate the Occurrence of multicentric cancers with a large number of pathologic and some clinical features disclosed a statistically significant association between multicentricity and grossly nonencapsulated dominant cancers with maximum diameters greater than 5 cm, the presence of a moderate or marked intraductal component and noninvasive cancer in its vicinity, and tumor involvement of the nipple. I n addition, it was noted that there was a greater likelihood that the primary tumor was of the lobular invasive type and that the overlying skin was involved when the multicentric cancer was invasive rather than noninvasive. Lymphatic tumor emboli were observed in quadrants in 18 or 2.0% of the cases. Although the number of examples is small, nevertheless positive associations were noted with the occurrence of primary tumors that were in the left breast or beneath the nipple and were not grossly circumscribed, but exhibited a nuclear grade of 1, intralymphatic and blood vessel invasion, calcium, and involvement of the overlying skin as well as nipple. I n addition, patients with such intralymphatic extension were more likely to have clinically detectable lymph nodes of which four or more contained metastases. T h e relationship of these findings to the rationale of such procedures as segmental resection in the surgical treatment of breast cancer is discussed.