Background: Vit D supplementation is common in BC. Given reports that Vit D modulates aromatase activity, we examined the relationship between Vit D and estrogen levels in PM BC. Methods: 125 PM women within 2 years of diagnosis of T1-3, N0-3, M0 BC were recruited in Toronto and Los Angeles between Mar 2009 and Jan 2010. Data included: anthropometrics (measured); medications and Vit D supplementation (structured interview); and tumor/treatment variables (clinical records). Serum was analyzed for 25(OH)D by radioimmunoassay (RIA) and estradiol (E2) and estrone (E1) by extraction-column chromatography-RIA. Statistical analyses used t-tests, Pearson correlation (r) and regression models after suitable transformation of variables. Results: Hormone therapies included AIs (n=56), tamoxifen (n=30) and none (n=39). The tamoxifen and no hormone groups did not differ for important variables, and were combined (non-AI, n=69). Mean age was 62 years, BMI 27 kg/m², Vit D supplementation 1244 IU/day, 25(OH)D 98 nmol/L; these attributes were similar in AI users and non-users. Mean E2 and E1 levels (pmol/L) were significantly lower in AI users (19.9 and 26.2) than non-users (39.6 and 123.7), both p<0.0001. The relationship between 25(OH)D and E2 differed qualitatively in AI users vs non-users (interaction p = 0.002): in AI users, 25(OH)D was positively correlated with E2 (r = + 0.31, p=0.02); in non-AI users this correlation was negative (r = - 0.26, p=0.03). 25(OH)D was not associated with E1 in either group. BMI was not associated with either E1 or E2 in AI-users (r = +0.09, p=0.52; r = - 0.03, p=0.84), but was univariably associated with E1 but not E2 in non-AI users (r = +0.44, p=0.0001, and r = +0.22, p=0.07). In multivariable analyses [age, BMI, 25(OH)D], 25(OH)D was positively associated with E2 in AI users (p=0.04) and negatively in non-AI users (p=0.05). BMI was associated with E1 in non-AI users (p=0.002). Measurement of AI levels is underway. Conclusions: In PM BC survivors, serum 25(OH)D was associated with higher E2 among AI users but not non-AI users. If replicated, this finding has implications for BC survivors receiving AIs. Funded by the Breast Cancer Research Foundation.