Breast carcinoma presenting with axillary metastases and no clinically apparent primary tumor in the breast is an uncommon form of stage II disease. Published studies have been characterized by small numbers and/or limited follow-up information. Although these patients are often looked on individually as having advanced disease, several published reports suggest that their prognosis is not exceptionally grave. The present study evaluated the largest series thus far described, consisting of 48 patients with a median follow-up of 5 years. Each presented with an axillary mass which proved to be metastatic adenocarcinoma, consistent with mammary origin when examined histologically. No patient had a palpable breast tumor. Mammography was negative in 28 patients (76%), and suspicious or positive in nine (24%). Nine (35%) of 26 metastases were positive for estrogen (ER) and progesterone (PR) receptors, 10 (38%) were negative for both receptors, and seven (27%) were ER positive/PR negative. Primary treatment was mastectomy and axillary dissection in 38 cases, 21 with adjuvant chemotherapy. A primary tumor was found pathologically in the breast in 36 cases (75%). Among 34 reviewed primary lesions, 27 (79%) were invasive and seven (21%) were histologically "noninvasive." Measured size was 0.1 to 6.5 cm (median, 1.5 cm). The number of involved lymph nodes was one to 65 with 20 cases having one to three positive and 20 having four or more positive. Follow-up ranged from 5 to 267 months (median, 60 months). Overall, 29 patients (60%) remained alive and disease free; two (4%) were disease free, but died of other causes; and the status of two (4%) was not known. Fifteen patients developed recurrent carcinoma, including 12 (25%) who died of disease. When compared with a matched series of stage II patients with equivalent extent of disease who presented with palpable breast tumors, patients with occult lesions had a more favorable prognosis overall, as well as when stratified by tumor size and nodal status, but the differences were not statistically significant. These results probably reflect the fact that the majority of stage II patients with clinically occult breast carcinoma usually have a grossly measurable invasive tumor detected pathologically. Consequently, the actual pathologic stage, which takes tumor size into consideration, determines prognosis rather than the apparent clinical stage described when the patient is first examined.