The inconsistent associations between fruit and vegetable intake and breast cancer risk may be due to heterogeneity of associations by estrogen (ER) and progesterone receptor (PR) status of the tumors. We evaluated this hypothesis in a large (2,386 cases and 2,503 controls) population-based case-control study in Poland, conducted between 2000 and 2003. We observed significant associations between reduced overall risk of breast cancer and increasing levels of total fruit intake (odds ratio (OR) for highest versus lowest quartile = 0.76, 95%CI = 0.63-0.91; p-trend = 0.01), but not for total vegetable intake (1.13 (0.93-1.37), p-trend = 0.25), after controlling for age, energy intake and known risk factors for breast cancer. The inverse association with total fruit intake was stronger for risk of ER+ (0.69 (0.54-0.88), p-trend = 0.01) than ER- tumors (0.89 (0.67-1.19), p-trend = 0.57) (p-heterogeneity = 0.02). In conclusion, this study suggests that fruit intake might have differential associations for breast tumor subtypes defined by ER status.