The unpredictable behavior of histologic types and the disputable results of particular surgical procedures have contributed to the controversies surrounding phyllode tumors. The therapeutic issue is further compounded by an imprecise preoperative diagnosis. This is illustrated in our series of 106 patients, in whom 71.7% were mistakenly operated on for fibroadenoma. Since only 15.8% of this group developed local recurrence, it may be acceptable to adopt a watchful policy, unless detailed microscopic assessment of tumor margins provides evidence of inadvertent transection. Reoperation would then be logical advice, as with one of our patients. In patients clinically suspected of having phyllode tumors, tissue diagnosis is preferable to investigative procedures, which are seldom definitive. Our experience has been with frozen-section diagnosis, although trucut core biopsy is certainly an attractive procedure. In 18 patients only 50% were positively identified as having phyllode tumors. The logistic and therapeutic issues related to use of frozen-section diagnosis are discussed. Wide excision has been recommended by many investigators as an appropriate measure, as long as it is technically possible to remove the lesion with an adequate rim of normal breast tissue. This applies especially to young female patients (less than 20 years), in whom conservation of the breast is particularly desirable and whose tumors behave in a more benign manner. On the other hand, mastectomy is necessary if the phyllode tumor is large compared with breast morphologic size,when histologic transformation has taken place, or when clinical behavior takes a more aggressive course. In this retrospective study, conservative surgery is generally practiced for both primary and recurrent tumors.(ABSTRACT TRUNCATED AT 250 WORDS)