PURPOSE: In this pilot study of high-dose-rate brachytherapy to the lumpectomy site as the sole radiation, ipsilateral and contralateral breast recurrences are documented with specific attention to the location of recurrence relative to the lumpectomy site. METHODS: Between March 1992 and January 1996, 39 patients with T1 (32 patients) and T2 breast cancers received 37.2 Gy in 10 fractions (b.i.d.) over 1 week prescribed to a volume encompassing the surgical clips. Thirteen received adjuvant tamoxifen, and 4 received chemotherapy. Follow-up included annual bilateral mammograms and clinical breast examination every 3 to 6 months. Whereas 13 patients had intraoperative implantation of the lumpectomy site, 26 had postoperative implantation. The latter group and 7 of the former group had surgical clips marking the lumpectomy site, which allowed estimates of the distance of any ipsilateral breast recurrence from the lumpectomy site, using the mediolateral and cranio-caudad mammographic views. RESULTS: At a median follow-up of 91 months, 33 women are alive, 4 have died of disease, and 2 have died of other causes. The 5-year actuarial rate of ipsilateral breast recurrence was 16.2%. Of 6 ipsilateral recurrences, 2 occurred within the lumpectomy site (in-field recurrences). One of the 2 patients had a 1-mm microscopic margin at initial diagnosis; the recurrence was a 3.5-mm microscopic focus of duct carcinoma in situ. The other patient had a 1.5-cm, high-grade infiltrating mammary carcinoma with no residual at wider resection at first diagnosis; the 5-mm invasive recurrence was also of high grade. Four women developed invasive recurrences at least 1.6 cm or more from the lumpectomy site (out-of-field recurrences). Two of these women had gross multifocal recurrences with two cancers in each patient; 1 of the 2 patients had an extensive intraductal component at initial diagnosis. The estimated nearest distances between the out-of-field recurrences and the surgical clips were 1.6, 5.5, 7.7, and 12.0 cm. All ipsilateral breast recurrences were salvaged by mastectomy (4 patients) or by repeat lumpectomy (2 patients) and whole-breast radiation. The interval postdiagnosis to ipsilateral recurrence ranged from 20 months to 58 months. There were two contralateral breast recurrences at intervals of 34 and 36 months; 1 of these patients also had a multifocal, ipsilateral recurrence at 58 months, as previously described. Among patients with any breast recurrence, 1 patient had a family history of prostate cancer; there was no family history of breast or ovarian cancer. Of 17 patients who received adjuvant systemic therapy, only 1 had a breast recurrence. CONCLUSIONS: In this pilot study, breast recurrences outside of the lumpectomy site were the predominant pattern of recurrence.