Sixty-six confirmed instances of clinically metachronous second breast cancers were encountered over a 10-year period in 1578 women with invasive breast cancer enrolled in Protocol 4 of the National Surgical Adjuvant Breast Project (NSABP). Seven of the second cancers were in situ, yielding an incidence of 3.7% invasive and 0.5% noninvasive cancers. Except for a peak of 1.75% in the second postoperative year, the annual incidence based on patients at risk was constant and less than 1%. Ninety-three percent of the second cancers occurred within 7 years and 80% within 5 years following mastectomy for the initial primary. All of the second cancers were regarded as being primarily of mammary origin exhibitng either: (1) an in situ component; (2) dissimilar but well-recognized patterns of primary breast cancers; or (3) the appearance of scar cancer, a recently described morphologic feature characteristically observed in some primary breast cancers. A search for factors that might be predictive of bilaterality was performed. Thirty-eight pathologic and eight clinical factors were assessed, including family history. Although the latter was 1.5 to 2 times more frequent in patients with bilateral disease, this estimate was not statistically significant. On the other hand, a statistically significant association with bilateral disease was found when the initial tumor measured more than 2.0 cm, was associated with invasive cancer or proliferative fibrocystic disease, nipple involvement, absent nodal sinus histiocytosis, lobular carcinoma in situ in the vicinity of the dominant mass, or was of the lobular invasive or tubular types. However, the degree of risk of these discriminants was no greater than 2 to 3:1. Despite the clinical scrutiny imposed by the NSABP protocol second tumors measured only 1 cm less than the first, measuring on average 2.4 cm, which reflects the difficulty attendant on the clinical detection of so-called early breast cancers. Yet, no significant difference in pathologic nodal status was noted between the first and second cancers or that of patients with unilateral disease. Furthermore, there was no difference in survival rate between patients who developed a second cancer and those with unilateral disease. These data, when considered with those of previous reports, indicate that: (1) the incidence as well as annual accrual per patients at risk of bilateral cancers is low; (2) most second cancers, although clinically metachronous, are probably biologically synchronous; (3) a second cancer does not influence the expected survival of patients who develop such an event; (4) the incidence of contralateral metastases from breast cancer is extremely rare and their distinction from a second primary depends principally if not exclusively on histologic assessment rather than on the somewhat empirical considerations often used in the past; (5) the influence of predictive characteristics for the development of bilateral disease that have been enumerated are relatively weak; and (6) it is justifiable to follow the contralateral breast of patients with breast cancer by careful physical and mammographic examinations rather than contralateral biopsies or mastectomies.