e24052 Background: We conducted an exercise study to IMPROVE health outcomes in older breast cancer survivors (BCS) from diverse racial and socioeconomic backgrounds. Here, we report cohort recruitment strategies and baseline characteristics. Methods: ‘IMPROVE’ is a randomized trial, designed to evaluate a group-based exercise intervention compared to support group. Participants were ≥ 65 years, had stage I-III breast cancer and within five years of treatment completion. Enrollment target was 220 participants, 25% in each of four strata defined by race (AA versus Non-Hispanic Whites [NHW]) and SES (low vs. high). Participants were recruited through hospital-based tumor registries, the state tumor registry or direct referrals. Results: Between October 2016 and November 2019, 7487 patients were screened, 4790 were potentially eligible, 230 were consented and 213 randomized into the study. The Eligible-to-Randomization rates were 4.4% overall, and 84%, 8%, and 2% for recruitment utilizing direct referrals, hospital and state registries, respectively. Median age of the randomized cohort was 70 years (range: 65-88 years) and included 25% AA/Low SES, 20% AA/High SES, 19% NHW/Low SES and 36% NHW/High SES Older BCS. Compared with registry-eligible patients, directly referred patients were more likely to be AA vs. NHW (41% vs. 19%, p = 0.003) and to accept study participation (88% vs. 16%, p < 0.0001). Conclusions: Direct referrals resulted in the highest recruitment yield and was the most effective strategy for recruiting AA patients. Behavioral intervention studies seeking to target older BCS from racial minority and SES disadvantaged backgrounds should focus on strategies that foster direct referrals to study participation. Clinical trial information: NCT02763228 .