Background: We investigated risk factor patterns for subtypes of breast cancer characterized by joint estrogen receptor (ER) and progesterone receptor (PR) status in a hospital-based case-control study. Methods: ER and PR tumor status were determined immunohisotchemically. Risk factors of interest were entered into a multiple polychotomous logistic regression model simultaneously; odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. Using this model, cases in the four tumor subtypes (ER + PR + , ER − PR − , ER + PR − , ER − PR + ) were compared simultaneously to controls. A Wald test for heterogeneity across the four subtypes was conducted, as well as a case–case comparison between the two most biologically disparate subtypes, ER + PR + and ER − PR − . Results: The receptor status distribution was as follows: 33% ER + PR + , 34% ER − PR − , 20% ER + PR − , and 13% ER − PR + . Among 317 cases and 401 controls, we found significant heterogeneity across the four tumor subtypes for older age at first full-term pregnancy (p=0.04) and post-menopausal status (p=0.04). For older age at first full-term pregnancy, an elevated risk was found for the ER + PR − subtype (OR=2.5; 95% CI: 1.2–5.1). For post-menopausal status, elevated risks were found for both the ER + PR + (OR=2.4; 95% CI: 1.1–4.9) and ER + PR − (OR=7.2; 95% CI: 2.4–21.7) subtypes. From the case–case comparisons, we found that cases, who had consumed alcohol for more than 1 year were 3.4 times more likely to have ER + PR + tumors than ER − PR − tumors (95% CI: 1.4–8.4). Conclusions: Certain breast cancer risk factors may vary by ER and PR status, and joint ER/PR status should be taken into account in future studies of risk factor estimates.