We evaluated the association between organochlorines and breast cancer subtype defined by the tumor characteristics: estrogen receptor status, progesterone receptor status, tumor size, and grade. A case-control study was conducted from 1995 to 1997 in Kingston and Toronto, Canada. Breast adipose tissue, taken from 217 cases and 213 biopsy controls frequency-matched on age, was analysed for 14 polychlorinated biphenyl (PCB) congeners and 10 pesticides. Adjusting for age, geometric means of several organochlorines differed by estrogen receptor status and tumor grade (p < 0.05). Odds ratios (ORs) for each organochlorine relative to the common control group for breast cancers of differing subtype were compared using polytomous logistic regression. Although the ORs did not differ significantly by subtype, the ORs of PCBs and p, p'-1,1-dichloro-2,2-bis(p-chlorophenyl)ethylene (DDE) were higher with risk of estrogen receptor-negative breast cancer than estrogen receptor-positive breast cancer. One of the most extreme differences was with DDE, where the OR for the association with risk of estrogen receptor-negative breast cancer was 2.4 (95% confidence interval (CI) 1.0-5.4) in the uppermost tertile relative to the lowest, whereas the corresponding OR for risk of estrogen receptor-positive breast cancer was 1.1 (95% CI 0.6-1.9). PCBs also tended to be more strongly positively associated with risk of larger and higher-grade tumors. The association between organochlorines and breast cancer risk did not significantly differ by subtype, but many PCBs were more strongly associated with tumors of poor prognosis.