When breast-conserving therapy was introduced at the Cancer Institute Hospital (CIH) in Tokyo in 1986, we instituted our own strategy as follows: 1) every effort is to be made for complete tumor resection while avoiding deformity of the breast, and 2) radiotherapy (RT) is applied only to the patients with positive surgical margins. This is, in turn, to clarify the group of patients in whom postoperative RT can be safely spared. Among 9670 patients operated on for primary breast cancer during the 16.5 year period from 1986 to 2002 at CIH, there were 2449 patients who underwent breast-conserving surgery (BCS). During the 6.5 years mean follow-up period, ipsilateral intrabreast tumor recurrence (IBTR) developed in 99 of the 2449 patients, with an overall rate of 4.0% and an annual rate of 0.62%. These 2449 patients were categorized into four subgroups according to either negative or positive margins and with or without radiotherapy. The IBTR rates and the number of patients in each subgroup were 5.5% in 1351 margin(-)RT(-) patients, 1.0% in 307 margin(-)RT(+) patients, 2.4% in 680 margin(+)RT(+) patients, and 4.5% in 111 margin(+)RT(-) patients. These results either with or without RT seem to be quite comparable to or even better than the results of BCS with RT reported from Western countries, where less emphasis seems to be placed on completeness of the local tumor resection with BCS, while RT is administered to basically all patients following BCS. IBTR was categorized into true recurrence (TR) and second primary lesion (SP) according to the margin status at the time of BCS, the former being lesions developed in patients with positive margins and the latter being those in patients with negative margins. It was demonstrated that in patients with positive margins, TR was much more common than SP, whereas in patients with negative margins, these incidences were just the opposite (i.e., TR was 60% less common than SP) and postoperative RT was effective in preventing both TR and SP, the effect on the latter being much more striking. With RT, the incidence of developing TR in patients who had positive margins was reduced to almost equal to that in margin(-) patients treated with no RT. Our method of IBTR categorization is based on biological consideration and detailed histopathologic examination, and appears to be the only biologically reasonable means so far that has been proposed for distinction between these two biologically different entities. TR and SP can be further reduced to exceptionally low levels in patients who received RT despite negative margins, though it would not seem reasonable to administer RT to all of these patients because the actual number of patients who would benefit is comparatively small. From these observations, it seems that our imaging, pathologic examination, and surgical approaches for patients who are candidates for BCS have been highly valid, and our criteria for sparing postoperative RT as well as categorization of IBTR into TR and SP are quite appropriate. Although our results with BCS seem to deserve wide recognition, they are not from randomized clinical trials, so the findings must be confirmed by a study in order to investigate whether the results at CIH can be applied generally at other institutions.