6550 Background: Physicians’ prior experiences caring for patients with breast cancer along with experiences in their social networks including family members and friends may be a key and understudied driver of recommendations for cancer screening. Methods: The Breast Cancer Social Networks study (CanSNET) is a national, mailed survey of 2,000 primary care providers (PCPs) randomly selected from the American Medical Association Masterfile. PCPs were asked to provide detailed characteristics on up to 2 women they know who have been diagnosed with breast cancer and “whose cancer, broadly speaking, had the greatest impact” on them, including friends, family members and patients. Each woman was categorized as being diagnosed (a) through screening with a good prognosis, (b) not through screening with a good prognosis, (c) through screening with a poor prognosis or (d) not through screening with a poor prognosis. We used a logistic regression model to assess the association between the network member and recommendations for routine screening mammograms to average-risk women ages 75+, adjusting for provider and practice characteristics. Results: Overall 871 physicians responded to the survey yielding an adjusted response rate of 52.3% (out of 1665 eligible). We found that 67% of physicians recommended screening for women 75+. The sample reported on 762 patients, 378 family members and 476 other network members who had been diagnosed with breast cancer. Ten percent of patients and 25.1% of family members reported on died of their disease. In adjusted models, we found that physicians who reported on family members who did not receive a mammogram and had a poor prognosis were significantly more likely to recommend screening compared to those who did not (Odds Ratio 1.22, 95% Confidence Interval 1.03, 1.43). Conclusions: Physicians’ experiences with their social networks was linked to their breast cancer screening recommendations, underscoring the potential for information that is learned from social networks to differ from clinical guidelines and highlighting the need to address a broad array of influences in trying to reduce potential over-screening in cancer.