The optimal treatment for patients with extensive pure ductal carcinoma in situ (DCIS) ≥4 cm is controversial. This study evaluates local relapse according to type of local therapy: mastectomy, breast-conserving surgery (BCS) alone, and BCS + radiation therapy (RT). Subjects were female patients who received diagnoses of pure DCIS ≥4 cm between 1989 and 2010 and were referred to British Columbia Cancer. Clinicopathologic and treatment characteristics were compared between treatment cohorts. Local relapse (LR) was estimated using competing risk analysis. Multivariable analysis was performed using Cox regression analysis. Patients had the following treatments: 490 mastectomy, 38 BCS alone, and 192 BCS + RT. The 10-year cumulative incidence of LR was 16% after BCS (95% confidence interval [CI], 6-29%), 8% after BCS + RT (95% CI, 4-12%), and 2% after mastectomy (95% CI, 1-4%). On multivariable analysis, estrogen receptor-negative disease (hazard ratio [HR], 3.32; 95% CI, 1.08-10.18; P = .04) and positive margins (HR, 3.55; 95% CI, 1.56-8.05; P = .002) were associated with increased LR. BCS alone (HR, 7.87; 95% CI, 2.82-21.92; P < .0001), BCS + RT + no boost (HR, 3.80; 95% CI, 1.56-9.28; P = .003), and BCS + RT + boost (HR, 5.76; 95% CI, 2.59-12.83; P < .0001) were all associated with a higher risk of relapse relative to mastectomy. Mastectomy remains a standard local treatment option for extensive DCIS, but BCS + RT may also be reasonably considered in selected patients with a careful discussion of the benefits, side effects, and patient preferences.