Background: Musculoskeletal symptoms, for which there is no standard therapy, are common in women on aromatase inhibitors (AIs) for PEBC. A phase III randomized study explored whether exercise improves arthralgias / myalgias (A/M). Methods: Eligible women with EPBC had A/M related to A and were randomized to an exercise (E) or a control (C) group. E patients exercised 3x weekly for 48 weeks; for the first 12 weeks in a supervised setting; for second 12 weeks, supervised once weekly and twice independently, and for the last 24 weeks independently, using aerobic and resistance programs tailored to their fitness. The primary endpoint was change in SF36 bodily pain domain scores at week 12 (W12). Secondary endpoints included change in bone density (BMD), change in body mass (BMI), strength, and in hot flash (HF) index. Results: The study closed due to poor accrual after 3 years with 22 (11C 11E) of the planned 72 subjects enrolled at two sites among 98 screened. Among 20 evaluable subjects, baseline median age was 62; BMI was 26 kg/m2 in the E and C arms; median number of A/M sites was 5, with a median worst score of 2 (CTC-version 2 criteria). In the E and C arms, respectively, mean baseline SF36 physical domain scores were 7 and 8 (best possible score 12, worst 2). At W12 the score improved in 6 (55%) E and 7 (64%) C subjects. Median weight dropped 4kg in the E and rose 4kg in the C groups by W48. W12 bench press and leg strength did not significantly change in either groups. At W48, there was minimal change in BMD in both groups compared to baseline, with only 1 C subject evolving from osteopenia to osteoporosis. Median baseline HF scores were low in both groups and did not differ significantly at any time point. Conclusions: A 48 week supervised exercise intervention did not have a measurable improvement in A related A/M symptoms as measued by 12W SF36 bodily pain scores. Weight dropped in the E group by a median of 4kg (8kg difference compared to C group). Recruitment is an important consideration in exercise trials and may be challenging among patients more distant from their initial diagnosis of cancer.