Purpose: Ductal carcinoma in situ (DCIS) is treated to prevent progression to invasive breast cancer. Yet, most lesions will never progress, implying that overtreatment exists. Therefore, we aimed to identify factors distinguishing harmless from potentially hazardous DCIS using a nested case–control study. Experimental Design: We conducted a case–control study nested in a population-based cohort of patients with DCIS treated with breast-conserving surgery (BCS) alone ( N = 2,658) between 1989 and 2005. We compared clinical, pathologic, and IHC DCIS characteristics of 200 women who subsequently developed ipsilateral invasive breast cancer (iIBC; cases) and 474 women who did not (controls), in a matched setting. Median follow-up time was 12.0 years (interquartile range, 9.0–15.3). Conditional logistic regression models were used to assess associations of various factors with subsequent iIBC risk after primary DCIS. Results: High COX-2 protein expression showed the strongest association with subsequent iIBC [OR = 2.97; 95% confidence interval (95% CI), 1.72–5.10]. In addition, HER2 overexpression (OR = 1.56; 95% CI, 1.05–2.31) and presence of periductal fibrosis (OR = 1.44; 95% CI, 1.01–2.06) were associated with subsequent iIBC risk. Patients with HER2 + /COX-2 high DCIS had a 4-fold higher risk of subsequent iIBC (vs. HER2 – /COX-2 low DCIS), and an estimated 22.8% cumulative risk of developing subsequent iIBC at 15 years. Conclusions: With this unbiased study design and representative group of patients with DCIS treated by BCS alone, COX-2, HER2, and periductal fibrosis were revealed as promising markers predicting progression of DCIS into iIBC. Validation will be done in independent datasets. Ultimately, this will aid individual risk stratification of women with primary DCIS. Clin Cancer Res; 24(15); 3593–601. ©2018 AACR .