To assess the significance of patient age, race, tumor-related prognostic parameters, status of surgical excision margins, and irradiation boost on incidence of ipsilateral breast relapse, and to review current issues in the management of T1-T2 breast cancer patients with conservation therapy. Records of 1037 patients with histologically confirmed stage T1 and 308 patients with T2 carcinoma of the breast treated with breast conservation therapy from January 1970 through December 1997 were prospectively registered and evaluated. The mean follow-up for surviving patients was 6.6 years (range, 4-30 years), with a minimum follow up of 4 years for all patients. There were 78 ipsilateral breast relapses (IBRs); the actuarial 10-year incidence of IBR was 7% for T1 and 11% for T2 tumors. In patients 40 years of age or younger, four of 24 (17%) with extensive intraductal component developed an ipsilateral breast relapse, compared with six of 80 (8%) without extensive intraductal component, in contrast to eight of 159 (5%) and 33 of 776 (4%) in postmenopausal patients with or without extensive intraductal component, respectively. In patients with T2 tumors, two of eight (25%) women 40 years or younger with extensive intraductal component, and seven of 50 (14%) without extensive intraductal component developed ipsilateral breast relapse. The corresponding values for the patients older than 40 years were five of 48 (10%) and 13 of 202 (6%), respectively. The incidences of ipsilateral breast relapses, correlated with status of surgical margins after re-excision in T1 tumors, were one of 30 (3.3%) for positive, no relapses in 40 patients with close margins, 16 of 438 (3.6%) for negative, and 18 of 196 (9%) for undetermined margins. In the patients with T2 tumors, ipsilateral breast relapses occurred in two of 16 patients (12.5%) with positive margins, one of 16 (6%) with close, seven of 105 (6.6%)with negative, and four of 68 (5.9%) with undetermined margins (differences not statistically significant). In patients with T1 tumors, negative margins, the 10-year relapse rate was the same (8%) in 559 to whom a boost was administered and in 66 without a boost. In patients with positive margins, the relapse rate was 4% in 215 receiving a boost (18-20 Gy) and 33% (two of six) without a boost. In patients with T2 tumors and negative margins, the rate of ipsilateral breast relapses in 16 patients to whom no boost was given was 12%, as opposed to 10% in 143 patients who received a boost. However, with T2 tumors and close or positive margins, the IBR rate at 10 years was 12% in 81 given a boost, in contrast to 40% (2 of 5) without a boost. In T1 tumors, the breast failure rate was two of 53 (3.7%) in women < or = 40 years receiving chemotherapy and eight of 51 (15.6%) without chemotherapy. For T2 tumors, the corresponding values were seven of 39 (17%) and two of 19 (10.5%), respectively. In women 40 years or younger with T1 tumors receiving hormones or not, the ipsilateral breast relapse rate was two of 19 (10.5%) and eight of 85 (9.4%), respectively; in the older than 40 years group, the corresponding values were six of 377 (1.6%) and 35 of 558 (6.2%). In the patients with T2 tumors, ipsilateral breast relapse rates were not statistically different in the various groups. On multivariate analysis, only age and adjuvant therapy were significant factors predictive of ipsilateral breast relapse. Surgical excision margins status following adequate doses of radiation therapy was not a predictor of ipsilateral breast relapse. In patients younger than 40 years of age with extensive intraductal component, a somewhat higher breast relapse rate was noted but not enough to preclude breast conservation therapy. A boost of irradiation did not have a significant impact in the incidence of ipsilateral breast relapse in patients with negative margins, but it was of benefit to those with close or positive margins. Close attention to surgical margin status and delivery of higher doses of irradiation to the tumor excision site in patients with close or positive surgical margins will decrease the probability of breast relapses.